This topic spawned from a AVS thread nominally about cloning some speakers geared towards prioritizing high sensitivity output capabilities as a primary requirement.

Main Reference Link: AVS Forum; How complex would it be to diy JTR copies?

“How Much Amplifier Power and Speaker Output?”

is a topic that often comes up, in many contexts, especially in DIY design when we’re selecting among components, hopefully taking their interaction into careful consideration towards a goal. We get to choose drivers. We get to choose amplifiers. We get to choose enclosure design. We get to choose the placement of loudspeakers and listening areas. If we’re really lucky and serious, we get to choose all aspects of the room itself.

Options are empowering, and if exercised intelligently, they are an efficient means toward end goals, if we understand consequences. That means understanding the interaction of factors, as well as the actual benefits of the end goal, spelled out into objective criteria. How much you need of anything depends pretty much on everything, from the source content itself, to your optimal personal playback preferences, and all things between. It is a complex question, with multi-factor answers.

As an end product, delivered to your lap, your head, at your ears, and ultimately between them, the end result of Sound Pressure Level, or SPL, is often related to ‘Reference Level’.

So Let’s talk about ‘Reference Level’,

In the aforementioned forum thread, Tom Vodhanel of Power Sound Audio had some really good points as to why this was not a trivial consideration, as did many others, and David Matthews shared this pretty good take in an editorial article.

I think this is particularly salient and a common discussion within DIY circles, as we often wonder about matching amps and speakers based on ‘power handling’ capabilities, sometimes to a dysfunctional level, followed by examining modeled output predictions to sometimes excruciating length. Chris Popovich might call it modelbating, an offshoot of measurebating, that can often become analysis paralysis.

While this happens often to nauseating detail when looking at predicted theoretical output, it often becomes very vague about how all of that translates to what we really get at the listening position, and how that compares to what we ‘need’ to achieve target playback capabilities.

What we actually need throughout the system, specifically, to satisfy our actual listening requirements, is itself a question that many don’t even bother to ask.

Many ‘System Designers’ are actual wizards, but more in the sense of configuration wizards that fill slots with options. For example, a client wants a quote, so they choose Widget A from Category B into Designated Quote Slot C, and keep adjusting choices in ‘D-Tools’ until that final total price of installed system quote adds up between Cost D minimum and Cost E maximum, with enough abstract brand and religion clout to feel good about it. I have watched it work that way, behind the scenes, and it reminds me of seeing how many clowns you can fit into a car to drive around in a circle. Most system ‘designs’ I have seen by companies selling ‘design and engineering’ are barely engineered, if at all. If they come with accurate wiring lists and connection diagrams that actually work, they’re better than average.

If we are actually doing system design engineering on a deeper level than having boxes that connect to make sound, we need to ask this question, among many, and we need to arrive at an answer, both in terms of end result capabilities, and specifically how we guarantee that we actually get there, with performance capabilities (beyond just SPL) intact: How Much Do I Need?

In depth interrogation of that question includes putting the screws into thumbs of the details, such as.

At your listening position?
From your loudspeakers?
From your power amplifiers?

Often, in real life, that question is answered not by self-evaluation, but by simply pulling a number commonly accepted by others. Of those times, most often that number is related to ‘Reference Level’. As in, “I want to be able to hit reference level”. Sometimes it may be, “I only listen 10-15 dB below reference, because beyond that it’s just too loud. ” Both of those answers are valid, though how one comes to that might benefit from some context.

Read it now, read it later. It’s a pleasant read.

‘Reference’ itself is a loaded word. It is often used in contexts such as ‘The Reference Standard’, as in a standard of the highest quality. However, the implications of the word, its history, and derivative nature of human beings, have created a gap between meaning and interpretation that is anything but consistent. There is a saying, I can explain it to you, but I can’t understand it for you. Reference level is exactly that, and I will attempt the former.

Manufacturer Perspective (Cody Hiebert). Reference level is interesting, because it shows no guarantee of sound quality at maximum output. What is the distortion metric on the speakers, where are the compression limits? Compression is the first indication of lost efficiency, and different distortions show how the speaker is deviating when suffering from efficiency losses. Without this information, Reference and associated headroom margin, or any other name you want to call SPL based cinema design, is just simply a number with no assurances attached. By its own nature SPL becomes ambiguous because in of itself it is a meaningless number with no real value. It does not guarantee linearity at all SPL points in audio playback.

There was a time Before Reference Level.

And it's not like we felt we were missing something. When I got into audio, surround sound at home was a pretty esoteric thing. Quadraphonic music had come and gone. A fine idea, but in actual practice, after the cost-benefit ratio played out, between more speakers and amplifiers, requirements of setup to yield meaningful benefit, largely nobody wanted it. I don’t know why, I was just a kid. As an older kid, I bought into DVD Audio, which also fell away due to lack of demand. Most people just didn’t care enough.

I kind of wonder how many people actually really listen into their music. I don’t wonder if most people do. They do not. I wonder about how freakishly small a minority we - ‘audiophiles’ who listen to music, only to listen to music, and savor every detail of sound events we can wrap our brain around - actually are. If you are not an audiophile, you just want to watch movies and hear what is going on, that’s fine. Sound quality still matters. For context, I am certainly an audiophile, perhaps to a dysfunctional degree, so we can say that disclosure on my perspective wrought through experience and temperament has been made.

Historical Context: From Dolby to THX

THX created "Reference Level" to match theater playback to the mixing studio, but listeners routinely turn it down for being too loud.

As far as standards go, back in my youth, it was kind of the epitome of the Spaghetti Western’s Wild west. Dolby was doing Dolby, effectively pushing Dolby Surround matrix encoding/decoding left/right phase differences into a mono two-speaker ‘surround’ component on existing 2-channel stereo formats. That eventually became known as Pro-Logic with an actively steered center channel.

Back when the dinosaurs of big fat full-range speakers ruled the earth (we just called them speakers), when it came to the question of “how much do I need?”, generally people either largely didn’t care, or they were still following the aftermath of the “Watt Wars” of the 70s, and the answer came down more to budget. You bought what you could afford, and if your priorities included loud, that meant big speakers, and big amplifiers.

And if we were talking honest power, back then the most cost-effective power-delivering topology that had any mainstream HiFi cred was largely Class AB amplification with linear power supplies, which meant Big Iron, big heat sinks, and if you had more than a couple hundred watts per channel via an outboard amplifier, people considered you a maniac. In college, I became that maniac, and none of my neighbors appreciated it.

As far as a ‘correct’ listening level goes, there was no such thing, and no one really cared. It was whatever you enjoyed, loud enough to hear everything, but not so loud that it became unpleasant. And as that level is subject to personal subjective experience, it made sense that people using audio systems at home for music or movies, either with Betamax or VHS tape, or for those who were super fancy, Laser Disc, in glorious CRT-rendered 480i, never bothered to even wonder what the playback level should be. Most of us had never heard about people using an SPL meter, let alone calibrating system playback. We ran it, and we liked it, or we bought something else.

And along came THX, a bearer of blessings that would make a nun blush, and among other things, more stuff to buy.

THX brought the notion of ‘Reference Level’, a standardization system intended and sold as a quality control program for commercial theaters by Lucas Films - THX was invented by audio scientist Tomlinson Holman at Lucasfilm in 1983.It was developed to ensure that the soundtrack for Star Wars: Episode VI - Return of the Jedi would be accurately reproduced in high-quality venues.

In commercial theaters, you cannot give everybody a ticket to their own remote. They’re all listening to the same thing, and they’re all going to have different preferences. So, as society demands sacrifices from individual freedom to work, this standardization component kind of made a whole lot of sense in commercial theaters. Pick a playback level that works well for most everybody, and stick to it, through the entire chain, from production to playback.

Who gets to decide playback levels? Well, leave it to the expert, the director! Or, at least leave it to those delegated by the director, and hope they make good choices.

The “Director's Intent" may be the most successful marketing phrase ever imported into consumer audio. The unforgiving truth is that most people won't sit through a film in the same room, on the same speakers, at the same level, under the same conditions as the mix engineer. How many of us have actually been in a dubbing theater and listened to the mix at reference level? Yet we're expected to treat a volume knob like a sacred covenant, leading us to aspirations of hitting “reference”, and it would seem that our future might become even louder.

In time, this standard was more miss than hit on a performance level. Despite this, THX Certification became one of the most recognizable badges in consumer audio, simultaneously serving as both an engineering standard and an extraordinarily effective marketing tool by boasting the clout and substance of cinema audio to consumer audio. There were some gains as far as standardizing minimum performance criteria, albeit tactically kept ambiguous from public view for pretty logical reasons, such as IP protection. There are a lot of personal anecdotes and professional assertions that indicate that a ‘THX correctly certified’ playback level is way too loud.

Reference Link: THX Blog THX Loudness

Reference Link: THX Reference

Reference Link: Acoustic Frontiers Reference Level Explained

People would complain in one direction, managers would turn it down based on the squeakiest wheel, and the standard of ‘Reference Level’ slid down the slope of ambiguity, unreliably, behind closed doors, in practice and due to popular demand.

Reference Link: SMPTE 2024 Media Technology Summit (Hollywood, CA)

THX built technologies around the reality that listeners routinely turn playback down. SMPTE researchers found theaters routinely operating below reference. Streaming platforms have made playback consistency more uncertain than ever. Yet the assumption persists that the standard itself remains beyond question.

A standard is supposed to describe reality. When reality routinely ignores the standard, there are only two possibilities: the public is wrong, or the standard is incomplete. One of those explanations is much more comfortable than the other.

And then THX came home, again in a mixed gift bag.

I want to be clear, I think THX offered a lot of good results, forcing an audio industry that largely disregarded many elements of performance in favor of ‘panache’ to pony up products that in some way prioritized peak output capabilities, lower distortion, flatish on-axis frequency response, relatively even horizontal dispersion from the front speakers, and reasonable low frequency audio system extension capabilities, and… SPL. A lot of those things, when the consumers got a taste for them, they really liked it, more than their breakfast cereal. I liked a lot of it. I still have some professional studio monitors that participated from the beginning in that program.

In a lot of ways, I think the THX program pushed the loudspeaker industry forward, at first kicking and screaming and then gleefully upon discovering the sales value of wearing that THX badge. I remember in a training class meeting a custom installer who LOVED THX systems, specifically because they were so easy to sell. ‘Reference Level’ came with that marketing maneuver.

Understanding Reference Level and Performance Standards

It is a calibration standard fixing peak limits at 105 dB for speakers and 115 dB for subwoofers to keep dialog clear while safely handling brief explosions.

Somewhere along the way, a calibration target became a virtue. Reference Level stopped being a tool and started becoming a sales sacrament. People weren't asking whether it served their needs anymore. They were becoming seduced by the promise of better by virtue of the Director's Intent.

In concept and original intent, I think it offers something potentially valuable, if all parameters required to actually execute it are met, from content production and delivery implementation, as well as system capabilities and setup, and a user wants to defer to the director (and the targeted understanding of the average audience needs) for the ‘correct’ playback level.

‘Reference Level’ playback capabilities were part of that, and assumed if you bought ‘certified’ equipment and did the very basic calibration of setting channel trim levels, and ran these systems in ‘normal’ home listening environments. After all, it was ‘Certified’, which kind of implies a promise.

A Promise of what?

For those that don’t already know, ‘Reference Level’ is nothing more than a system gain and output level standardization method both in content production, and system setup calibration.

At home, as with the original commercial THX program, ‘Reference Level’ potentially allows content producers, when implemented from beginning to end, to provide a means for end users to experience ‘ideal’ suggested playback levels for that content, where listeners can easily understand dialog and environmental/action sounds at the ‘average’ playback level, consistently, comfortably. It also allows enough ‘headroom’ above long-term steady playback content like voices or ambient sound for periodic transient louder content to safely and briefly exercise the top end of a pretty capable system’s dynamic range. Explosions, gun shots, falling objects, Godzilla coughing, etc.

These peaks operate within levels that, for a short duration, are still safe for listeners and most equipment. Even if that equipment fails to meet those output levels uncompromised, at least the equipment probably won’t break. Those peak levels that are safe as short-term occasional transients add some extra excitement and realism versus lower dynamic ceilings that would require either lower overall playback levels (with quieter dialog) or more electronic dynamic compression applied to the audio tracks, which removes the ‘live’ character of those events.

‘Reference Level’ establishes a ceiling to provide ‘headroom’ for the playback of content channels, such that 105 dB is the highest level available for a full scale maximum output sine wave for each of the ‘main’ channels, and the Low Frequency Effects channel, offset by 10 dB in the encoder and decoder, may be as high as 115 dB in peak delivered output, to supplement the ‘Boom Boom’ capabilities of playback without compromising the noise floor of the main channels at the time. Barry Ober called it the ‘Dinosaur Stomp’ channel. Barry Ober also had some interesting commentary regarding the origin of the LFE channel for commercial formats, and why using it at all in a home release when you had 24 bit formats and bass management didn’t actually make any sense, but that’s a topic for another day. I miss that guy.

Combined with dialog normalization, metadata that ‘should’ be included with the audio data, the processor will know how to ‘slide’ the output level of the content under the hood (effectively adjusting the master volume without changing the master volume display) so that regardless of where the dialog level sits relative to the highest peak value (which in digital formats is typically normalized to peak track output levels to maximize signal to noise ratio), dialog can be ‘anchored’ at comfortable and intelligible speaking levels, akin to if you were witnessing a conversation, somewhere in the range of 75-85 dB, and then the dynamic peaks land where they do relative to the target dialog levels.

Brian Florian, my dear Canadian Friend, did a nice article a long, long time ago explaining how dialog normalization works, and why.

Reference Link: Home Theater HiFi "Dialogue Normalization: Friend or Foe"

So, what this means is, if a system is properly calibrated so that ‘Reference Level’ actually IS at reference, and you’re listening to content that properly implements dialog normalization so that ‘Reference Level’ actually has any meaning, when you set the volume control to ‘Reference Level’, or ‘0 dB’ on the reference relative scale, IF the people producing the content adhered to ‘Reference Standards’, you may have playback peaks up to the peak voltage on an amplifier output that would be required to elicit 105 dB with a single sine wave, plus or minus variations due to loudspeakers and room response.

In the original reference thread, Tom Vodhanel brought up a valid piece of minutia, in that 105 dB ceiling relates to a sine wave signal, and that content isn’t that. It is in fact minutia, but I think it deserves a side quest.

The peak of the delivery format is fixed. There is no higher, and peaks are what get clipped when the power amplifier ‘clips’ as the output ‘tries’ to exceed the available rail voltage. But a sine wave RMS value is 0.707… of the peak value, and it is theoretically possible to sneak a bit more RMS content with mutli-frequency signals.

So, this discrepancy can be relevant if we wanted to just get extremely argumentative. We could point out that with a ‘perfect’ square wave (as perfect as the format sampling bandwidth limits will allow) at max signal voltage defined by 0 dBFSD (dB Full Scale Digital) digital format limits, the theoretical maximum SPL ceiling output would actually be 108 dB. A square wave packs odd order harmonics until theoretical Infinite Hz onto that sine wave in a way that keeps the peak the same as that sine wave alone, but the wave is as constant in peak output as the bandwidth will allow, so that where peak waveform amplitude and RMS amplitude are the same thing. As such the point is that it is theoretically possible to pack a content signal with more RMS value, and as such slightly exceed 105 dB, without increasing the peak voltage.

Having interrogated that possibility, practically speaking, I would release it from serious consideration. One could make that argument, sure but that square wave example would only happen in listening content through malicious malfeasance or extreme incompetence, and that would imply the WORST technically possible case scenario. 3 dB of playback headroom would guarantee plenty of margin, and margin nice, but exponentially expensive. 1 dB is probably more than enough, ever, with content. Find me an exception, and I’ll buy you a sandwich for your contribution. To be clear, I don’t think Tom was making the argument that you needed to add headroom because of that, I think he just mentioned it as a point of interest, and I thought it interesting enough to include.

But what you CAN count on is that the peak voltage levels of the electronic waveform will not exceed that required to reproduce 105 dB with a sine wave at the ‘nominal’ output level. By ‘nominal’ I mean that the loudspeaker output itself, even with constant input levels, is all over the place as you change frequency and direction, but somewhere you pick the ‘average’ as it is relevant to your use, and you ‘name’ that value. Within the parameters of what qualification, what that means is that for each main channel, if we are implementing ‘Reference Level’, then 105 dB, that’s the nominal ceiling, period. Whether you want to bump up your requirements in this hypothetical another 1-3 dB is entirely up to you.

But wait, there’s more, because we have complications. RP22 is recommending MOAR headroom, possibly with good reason, if you understand what those recommendations actually mean. Some of the details are unclear, or even undecided.

I say this because with the recent discussion of RP22 in that thread, Schurter was good enough to summarize a lot of talk about the value of ‘headroom’ beyond the actual listening requirements at ‘reference’ level.

After mentioning this reference, I bothered to read through RP22, and I feel compelled to say that it has a lot of good, generalized recommendations for thoughtful consideration, covering a broad swathe of relevant topics in audio reproduction. There is a whole lot to applaud, and I am figuratively applauding into my keyboard literally standing up, at my ‘standing desk’ that is actually a commandeered tall café table.

I also think that the way it is written, as is typical of publications written for audiences with less technically nuanced understandings of topics, is largely a very broad summary of, and an incomplete work at that, it leaves lots of room for misinterpretation. It is also not immune to industry influence that has “how do we capitalize on this?” built into cooperation requirements.

RP22: A New Era Of Uncertainty

New CEDIA tiers push for extreme, long-term loudness targets that tempt buyers into overpaying for massive volume while sacrificing actual low-level audio clarity.

While I’m not seeking any grand jury indictments nor am accusing anyone of making deliberately misleading statements, if you consider the tendencies of human beings, when it comes to “Recommended SPL output capabilities”, misinterpretation is kind of baked into our decoders.

These ‘engineering-oriented objective recommendation standards’ (I’m paraphrasing, but I think it’s a fair summary), could lead to a chain of errors due to misunderstanding and misinterpretation. Consider these possibilities and factors.

Recommendations include ‘Performance Levels’, 1-4, and target metrics that have been selected from among expert input on the topic. Typically experts disagree about many things, because they deal in nuance, personal experience, and have different perspectives. I would guess that experts who contributed to RP22 have some disagreements to some extent. And there is certainly remaining debate among experts within the industry at large. This is far from settled, or even completely spelled out in the document, because details matter.

In terms of SPL, reading off of the RP22 document on page 77, in this example screen speaker output capabilities delivered to the ‘Reference Seating Position’, “(post calibration EQ, within assigned bandwidth), without clipping. ” It provides a rough calculation example that basically factors in published voltage sensitivity, power rating, and inverse square law for ‘point source’ (as opposed to ‘line array’) propagation loss, the fall of SPL as a function of distance.

It is further clarified that these values are minimum long-term SPL capabilities, as per AES75-2022, or ANSI-CTA 2034-A, details not yet determined, nor specified how long qualifies as ‘long-term’…

It also suggests considering…

Additional speaker SPL capability at bass frequencies to allow for bass contours (for which there are no guidelines).

and

Additional speaker SPL capability to allow for +EQ (typically applied by ‘room correction, but could also be loudspeaker compensation, ‘transparent’ screen compensation’, or just part of generalized tone controls.

These are important additional considerations, which we’re going to crunch numbers later, but for reference and consideration, these are ‘recommended’ targets corresponding to ‘Performance Levels’ 1-4 for ‘screen speakers’(aka, LCR), before additional SPL capabilities are considered:

Level 1: 99 dB Minimum, 102 dB Recommended
Level 2: 102 dB Minimum, 105 dB Recommended
Level 3: 105 dB Minimum, 108 dB Recommended
Level 4: 108 dB Minimum, 111 dB Recommended

Subwoofers:

Level 1: 109 dB Minimum, 114 dB Recommended
Level 2: 112 dB Minimum, 117 dB Recommended
Level 3: 115 dB Minimum, 120 dB Recommended
Level 4: 118 dB Minimum, 123 dB Recommended

People who seek a simplified understanding to make decisions, which is most of us, especially end users who are not seasoned cynical veterans of the industry, gravitate as much as possible towards collapsing reality variables into simplified, seemingly objective metrics. Then they may mistake these generalized (and debatable) recommendations as stringent requirements, assuming that the set of ‘Performance Level’ criteria the system meets will be a critical predictor of actual use benefit. In our circles within Harbottle Audio, we are seeing evidence of this, which is a little concerning for reasons to be mentioned.

These ‘Performance Level’ criteria are by definition a collapsing of nuance, for the sake of simplifying decisions, even if they are fantastic for general guidelines, becoming inflexibly attached to these sets of targets may actually prove detrimental. I. e. , “I want Level 4 Performance” instead of “I want performance that provides my optimal experience. ”

In peak SPL requirements, there is nothing that competes on a performance to dollar ratio with cheap PA equipment. Taking it one step further into loudspeaker requirements as represented in RP22, in areas where some PA equipment might fall short, you can simply reshape horns, maybe swap in equally cheap smaller drivers to improve directivity, lean even MOAR on non-linear port output and smooth out amplitude response with electronic correction, and according to RP22 guidelines, so far, you’ve got an ideal speaker. In fact, if I was only concerned with meeting performance standards for delivering sound related to RP22 recommendations, this would be my recommended course of action. Why pay for all other aspects required for comprehensive sound quality if the main purpose of the system is to simply get to ‘Performance Level 4’?

Manufacturer Perspective (Cody Hiebert). In PA gear a lot of it does not compress or distort horribly… At their maximums. PA is designed to be beaten on for extended periods of time. However, they are also not fidelity machines either. The very reason they can endure the beating is because they have very stiff suspensions and they limit stroke through the motor topology on the drivers. The drawback is that the distortion they exhibit is not THD. It's generally IMD and Temporal distortions. So what happens when the signal deviates from the test signal averaging model, for example in the wormhole scene in Interstellar… which is like 75 seconds of pure insanity.

So when the appearance of a target is ‘loudspeaker maximum linear sound levels’, we have to ask what kind of linearity we are looking for. A system that is optimized for high SPL is not automatically optimized for low SPL just because it is designed to go loud. Race cars are designed to go fast and suck at going slow. So what are we giving up?

To be clear: This is not a swipe at high sensitivity loudspeakers that use compression drivers or horns. In fact, one of the BEST audio systems I’ve ever heard was a Magico horn-based system that used horns not just for tweeters and midrange reproduction, but woofers. I am pointing out the reality that implementing techniques leveraged in cheap PA gear to standards that meet requirements of high fidelity playback preclude cheap, because there is in fact a lot about high fidelity audio reproduction beyond just loud. Such examples of ‘PA Style’ loudspeakers in the HiFi market are often exceptional in that they don’t just sound good playing loud, but they also don’t blatantly sound like PA equipment. In the same way that in other HiFi speakers you don’t hear the metal domes, the ribbons, the electrostatic panels, the paper, or the carbon fiber. You hear the audio, largely only the audio.

What I am concerned about is mistaking and fixating on a generally fine performance recommendation SPL level which is great, all else being equal, and throwing a lot of other performance capabilities under the bus because of the fixation on maximum SPL capabilities under test conditions. What happens when the content deviates from the test signal?

Something also interesting about these recommendations is that for Levels 3 and 4, they are advocating “minimum ‘long-term” output levels beyond what properly executed ‘reference level’ would ever require.

RP22 does not provide context as to why, other than for “predictable, distortion-free headroom” (Page 39), especially since content, by nature of the design of ‘Reference Level’, does not provide long-term peak output, especially to main channels. That ceiling was intended for short-term bursts that don’t sound as loud as they physically are due to the nature of our auditory system. ‘Reference Level’ ceilings were intended for brief transients, largely unused. Depending on what RP22 means by ‘Long-Term’, it is not only FAR more taxing on equipment, but also potentially causes hearing loss if the sound quality isn’t of a particular caliber, which they admit on page 14 of the Reference Audio Level and SPL Capabilities whitepaper.

As far as actual high-end users are concerned, and I mean people investing hundreds of thousands of dollars into just the audio reproduction equipment before we even address the room or video systems, I have NEVER spoken to a person who, when listening to system and content reference level output, wished to play louder. If anything, reference level comes out for a few demos, and then in real life, they rarely if ever go near it. I’m not saying anybody should or shouldn’t listen below, at, or beyond ‘reference level’ as a matter of correctness. I think people should do what they want.

However, from a ‘performance level’ recommendation standpoint, which implies sound quality as is their stated goal on page one of the RP22 document, it is interesting because it implies designing and paying for more of what you’re not going to use is a worthwhile investment for most people, when in fact it is a fringe real use case. If somebody is trying to put together an audio system with budget constraints, and blindly shooting for SPL ‘requirements’ because they incorrectly assume that they are a valid ‘standard’, that will go poorly for those seeking fidelity, detail, and true ‘immersion’.

Don’t get me wrong, EQ considerations I think are important, however, the recommendations mention EQ lifts applied to be taken in consideration in addition to these recommended levels beyond ‘Reference Level’. MOAR ON MOAR.

Manufacturer Perspective (Cody Hiebert). Add to this, their statements about calculating the headroom needed to hit the SPL performance Level. Depending on the room design, they state typical +10dB for the LFE channel, then if you use bass management, there is another +4dB (approximate), now we need to add contouring, so if there is a +6 dB subwoofer contour, and we want Level 4:

123+10+4+6 = 143 dB at the RSP. This can include room and boundary effects as well. My concern is that they refer to 20 Hz to 120 Hz. If someone is chasing the SPL bragging rights, this is pretty heavy stuff. For the sake of headroom and linearity, compression and coherence limits in AES75 do not guarantee sound quality. Compression is the loss of efficiency, it’s not the descriptor of linearity in all things system related. As I have stated before and many times, it is an indication. Coherence is a broad catch all of all kinds of distortions, but it will not be able to tell you which ones are the problem, therefore, when the signal deviates from M-Noise the odds of the same failure point manifesting in the same way are very low. If the program material provides the right kind of stress, it bypasses the distributed averaging of M-Noise and aggressively targets a single physical vulnerability which can cause early distortions and compression to show up when the system technically passes AES75. Like the wormhole scene mentioned above.

In my opinion, CEDIA is trying to make a system for sound quality by using SPL at the worst case scenario as the metric. The issue is that those two things are not interchangeable and they do not express themselves the same way from an engineering perspective and in real use.

To be clear, depending on the equipment, there CAN be benefits to having more output than one might actually use, in that at maximum levels as defined by amplifier clipping, the speakers sound like utter garbage, and as you back down, things clean up, sometimes a bit, sometimes a LOT. This is, in fact, the nature of loudspeakers, generally speaking: the further you operate away from the point of destruction, the less the system will sound like it is going to destroy itself. Is this always the case? No. Some systems just sound like they want to die after 10 watts, and some give no indication of stress until way past that.

I am not arguing against the logic or reasoning of having more than you would use, and for those who see it and would point that out, point valid. RP22 didn’t point that out, but that doesn’t mean it doesn’t exist, or won't exist in future versions, only that I do not know the reasoning for those recommendation standards being lumped into a single index guide, ‘Performance Level’, when the thing that dictates usable sound is found outside of SPL.

Seven Frameworks for Understanding SPL Requirements

I would also like to build and expand on the reality of loudspeaker distortion typically falling with output, as it relates to the value of ‘headroom’ that will likely never see the point of stress, that the behavior in which loudspeakers do this carries a lot of nuance, and I have seen ‘headroom’ discussed as a rationale equating higher maximum SPL capability with better fidelity at any level. This is very clearly wrong.

With different loudspeakers, when you find a point at which they are equal, even ‘qualified’ maximum SPL capabilities cannot describe output quality, qualitatively.

For those interested in nuance, I think it’s worth spelling out. If you just want to believe that the ability to play loud means that it plays cleaner at any level than something that will not, abort, abort, abort!

If so, don’t let me waste any more of your time. Life is short, and you’re going to die one day. Do something more worthwhile. Watch a movie, hold hands, live, love, drink responsibly, or don’t.

I don’t think CEDIA is claiming that louder maximum output capabilities between devices automatically means better playback below those levels, at any level. That’s just TOO stupid to have the involvement of an industry that often puts real effort into delivering comprehensively good sound quality. Harman, for one, would never go along with that. Choose your favorite loudspeaker manufacturer that actually spends time and money on R&D. That probably includes them.

But sometimes people seem to believe this, and it is understandable why, because it is a product of logical, even if simplified, reasoning. If distortion falls with level, and a loudspeaker distorts less as you lower the level, then it will sound better as you back down away from ‘maximum’. Yes. The error in thinking is applying that between different devices, which have different distortion profiles, objectively, categorically, qualitatively, and the complexity of that variation is immense.

If system SPL delivery capability ‘headroom’ actually remains headroom, because things happened as designed, that means you don’t use it. In a perfect system, if you don’t use ‘headroom’ beyond the ‘Reference Level’ ceiling, the headroom gives you nothing directly, nada.

Framework 1: The Headroom Illusion

The Headroom Illusion is the belief that unused SPL capability automatically improves playback quality.

Headroom is insurance. Insurance is valuable. But insurance only creates direct performance benefits when it is actually used. If a system never approaches its output limits, additional unused output capability is not a sound quality upgrade—it is merely additional safety margin. Margin is good. Maybe you like being tall, and one day you feel like wearing heels. You never know. Better to have it and not need it, than need it and not have it, as they say. But you pay for that. Maybe better to pay for what you plausibly might use, rather than for what you likely never would even want to?

SPL is selling. Who’s Buying?

In every hobby there is a point where performance stops being the product and status becomes the product. In HT audio and private screening rooms, maximum SPL often sits right on that border. Nobody wants to admit it because numbers feel objective, but a lot of systems are purchased for the same reason people buy watches rated for depths they'll never dive to and sports cars they'll never drive flat out.

So, THX set up ‘Reference Level’ as a target of aspiration. It became, as with many things established by the THX program, a sales tool.

Now CEDIA might seem to be saying via RP22 that you should have even more output than that. And some people may interpret the higher ‘Performance Levels’ attaching higher recommended SPL capabilities with comprehensive sound quality capabilities. Standards have a funny habit of becoming aspirations, and aspirations have a funny habit of becoming insecurities. What begins as a recommendation quietly evolves into a hierarchy. Before long, people aren't asking whether a system sounds good. They're asking whether it qualifies.

In all honesty, I’m not really sure what they’re trying to imply with these output recommendations, because the current RP22 documentation doesn’t seem to explain the benefit of having more than you actually determine your ‘need’, aside from wanting a ‘Performance Level’ designation. Those recommendations are made without any specific explanation of why.

Getting there. Get a quote for the ticket before getting on that boat.

Manufacturer Perspective (Cody Hiebert). From my perspective in manufacturing, SPL is the generalized term for all kinds of auditory goals. Do you want more impact? Then you need more SPL. Do you want more immersion? Then you need more SPL. Rarely are the effects associated with dynamic capacity and complex loading capabilities, rather they are reduced to a set of SPL numbers assigned to a frequency range (or not assigned frequencies), and then equated to a driver diameter. This is generally how subwoofer sizing goes, even in professional integrator circles where “there is no replacement for displacement”. Simply because the verbiage to articulate what is actually happening has been literally thrown away by current measurement verification standards. Seriously. AES75 is the only one that gives you an indication of when problems arise, but it can’t tell you what they are or if they will be better or worse when the signal deviates from M Noise.

But the SPL recommendations are what they are, and if you want them, it would be good to understand what that requires. This is not a trivial thing that is going to be a cake walk for any quality product in any ‘normal’ listening environment.

It’s not a small thing. It’s not an easy thing. And if you’re going to achieve it by delivering good sound quality through the entire dynamic range, it’s not a cheap thing. In the context of really good sound quality, quality that most people can easily discern, you’re probably not going to actually pull it off with cheap PA equipment, even if tweaked for dispersion and frequency response requirements to qualify for what RP22 does spell out for loudspeaker behavior recommendations.

As Tom Vodhanel rightly pointed out, ‘Reference level’ is hard enough in a simple world.

For illustration, I’m going to do calculations using inverse square law, published (promised) loudspeaker voltage sensitivity, and voltage equivalent to rated power, similar to the RP22 example, but with more direct math, at different distances, to consider direct (unreflected) sound, with no room ‘help’.

Making the distinction of direct sound is important, for two reasons. One, it’s the only sound we can count on, assuming a clear line of sight. Two, it is the sound that isn’t jumbled in time by the room. Direct sound is the core foundation of ‘intelligibility’.

Reflected sound may up your SPL at distance, and may sound cool and spacious when carefully applied, even slightly more ‘present’ or ‘detailed’, if you get lucky or know how to manipulate it. But in the range of dialogue and up, where audible details get critical to movie enjoyment (dialogue is most often crucial to story), good, intact, faithfully reproduced direct sound is really what you’re predominantly after.

If you like your reflected sound components, I’m not suggesting you change that, just take it as a bonus, because it’s free in terms of generated output. A ‘Recycling Benefit’ of sorts.

My point is also consistent with the RP22 offered estimated calculations and also consistent with observations and discussion related to the article in Residential Systems that David Mathews shared, as it relates to situations with lower proportions of direct sound needing higher SPL for similar subjective effect. Above the ‘transition’ frequency, it is the direct sound that dominates the experience, and if you’re talking about fidelity, intelligibility, etc. , it is the direct sound that we want to dominate. RP22 recommendations in the acoustics section largely validate this, even if at times inconsistently.

At the lowest frequencies the room itself becomes part of the impulse launch, the immediate boundaries somewhat blend into the baffle itself, and the first reflections may combine with the first launch to change the impulse response, but more like a waveguide with a kind of haphazardly loaded driver than with distinct separate first arrivals.

As frequency drops, there is less ‘purely direct’ sound in the mix from a standpoint of the strictest definition. Fortunately, because the wavelengths make that distinction less distinctive, maybe we care less. In a way, the initial reflected sound from the first launch, because it recombines cohesively, maintains first arrival behavior, to a degree, changed in shape, but perhaps not significantly degraded in terms of ‘transient response’. Reflected sound isn’t necessarily reverberation, nor a resonance, both of which are low frequency detail killers. In low frequencies, reflections with short-delay times CAN preserve temporal, ‘transient’ detail, even if it may be not quite the same as the source heard without reflections.

That’s a heck of a spin off topic, but I think the quick of it is, we’re not talking about that. For very low frequencies produced by the subwoofers in a home setting, we aren’t talking simple point source inverse square law propagation loss relationships.

Acknowledging all that, we’re going to shelve it, for the time-being. Back to the simple, horrifying point, talking about the ‘Main’ speakers, and reliably meeting that ‘Reference level 105 dB’ output level ceiling capability to ensure ‘headroom’ in content does not become a bumped skull.

And now, a reality check: Calculating what we need for Main Speakers to reach… ‘Reference Level’.

Framework 2: The Reference Ceiling

Reference Level is best understood as a maximum design envelope rather than a mandatory listening target.

Manufacturer Perspective (Cody Hiebert). Reference Level establishes the maximum dynamic range a playback system should be capable of reproducing. It does not imply that every listener should routinely listen at that level, nor that every piece of content will require it for the simple reason that mixing studios and systems are standardized to "Reference" but then use about 20 to 30 db down from their design limit; which in simple math is using about 0.1% to 1% of the design, while maintaining the reality that this severe neutering does not actually guarantee perfect linearity at all points at all times. If RP22 is going to establish a worst case scenario as a basis for the headroom they are asking for, the tests and verification should likewise be the worst case. And nothing is more worst case than a full scale sine sweep.

For a ‘Main’ speaker, our ‘Reference Level’ peak Target is 105 dB as a sine wave. MUCH of a main speaker’s operating range will be above the ‘transition frequency’, wherever that may be in any given room, and it differs depending on the acoustics. But for our example, close enough, and it is an easy calculation to standardize.

In my calculator, we can predict the ‘power’ required (equivalent voltage at nominal impedance) given an input of a loudspeaker's voltage sensitivity, distance, and target output level, in our case 105 dB.

I’m going to grab specifications provided by a pretty mainstream manufacturer that look plausible for the loudspeaker, typical of many quality loudspeakers in the price range of $1,500 each to run our examples, a pretty ‘Full-range’ tower speaker, in terms of frequency response anyway. You infrasonic bass heads, keep the cackling down, I know, I know. If it’s not reaching easily into single digits, not really ‘full-range’. Most of the kids who didn’t eat the Play-Doh find it ‘full-range’ enough.

This is neither an endorsement nor indictment of the product. I just wanted to grab a real life example. These look cool, and lots of people know SVS as a reputable brand that represents reasonable or better performance results for the money spent. I am using it as an example because it is representative of a lot of very decent home audio loudspeakers.

FWIW, all 13 reviewers gave it 5 stars, so that’s worth something. More to the point, I’m pulling specifications that are both plausible as well as consistent with what the RP22 example was using to represent a predictor of target capabilities.

SVS Ultra Evolution Tower specifications — SVS Ultra Evolution Tower Specs

Framework 3: The Distance Tax

The Distance Tax is the exponential increase in system resources required to maintain the same direct sound level as listening distance increases.

Every additional meter increases the demands placed upon the loudspeaker, amplifier, thermal capacity, and distortion performance. Distance is not free. Distance is paid for in watts, displacement, distortion, or money.

So salient data to our simplistic calculator:

87 dB@2.83V@1Meter
Recommended amplifier power 20-300 watts

Let’s see how this ‘works’ with ‘reference level’ of 105 dB output at the main listening position. Main listening position requires distance values. Let’s look at a variety of options, in 1 meter increments, from 1 meter (near-field mixing monitor distance, and a standard for loudspeaker specifications) to 4 meters not a particularly extreme distance, and see what that requires of amplifier ‘power’ and associated loudspeaker ‘Power Handling’ (admittedly a slippery limit).

SVS Ultra Evolution tower distance to power calculations — SPL at Distance Inverse Square Calculations

So with these five star tower speakers, we can see that if you use them as nearfield mixing monitors, 1 meter away, from an SPL standpoint, we’re probably fine. 63 ‘watts’ of equivalent voltage at 8 ohms, or peak RMS voltage of 22.5 volts, or waveform peak voltage of 31.8 volts, any decent AVR should be able to swing that, at least a couple channels at a time. If the load dips to 4 ohms, that would equate to 126 watts/channel. Not insignificant, but most AVRs will probably not fall short by much, if they do. But most people are probably not using them at 1 meter distance.

2 meters is probably near the minimum plausible for a very small ‘theater’ listening space. Maybe good for a small couch and cuddle time while watching a movie. We now need 45 volts RMS, or 63.6 waveform peak voltage, equivalent to 253 watts at 8 ohms, and into the nominal 6 ohms, 337 watts, into 4 ohms if the speaker dips there, potentially, 505 watts. Uh, probably not the typical AVR. Probably not AVR power at all. Okay, well about that Certification Badge, how much did we pay extra for that? Don’t worry, just upgrade!

Enter your hero, the outboard power amplifier! If you want to do what you thought your AVR should be capable of doing all by itself, thankfully you can just buy another box. In all fairness, outboard power amps are a standard feature in any moderate dedicated audio space.

Thanks to the development in manufacturing of modular Class D amplifiers by Hypex, ICEPower, Purfi, and etc, there are a slew of reasonably affordable power amplifiers that have specifications that fit the theoretical requirements of this case. This is genuinely pretty cool for people who can now have significantly more output capability than what typically fits in an AVR, with reasonably more expenditure and rack space. Buckeye and VTV are forum favorites, but manufacturers that essentially buy in bulk, install in a chassis with connectors, give it a name and a special superiority story are popping up all over.

There are nuances to that assumption, in terms of whether the amplifiers can actually provide what the specifications imply without engaging protection circuits or ‘soft clipping’, over the entire operating range. But we’re shelving that, as we’re still in simplistic mode, and will just say they do.

Good to go, so long as your loudspeaker holds together. It is often said by audio salesmen (I used to be one) about any loudspeaker they are trying to sell, that “it can handle more power for short periods if it’s clean. ” That can be true. It usually is. In this case, the theoretical nominal power required at 6 ohms (337 ‘watts’) is just barely beyond the recommended power amplifier range, so if I was betting my lunch money, I would bet on survival.

Back that up to 3 meters, or just short of 10 feet.

Things just got serious.

Framework 4: Output Economics

Output Economics describes the rapidly escalating cost of achieving incremental increases in clean SPL.

As output targets rise, the costs are never linear. Every additional decibel demands disproportionately greater resources from the amplifier, loudspeaker, enclosure, cooling system, and room. The last few decibels are always the most expensive.

We now need an amplifier that can ‘peak burst’ (short duration) 67.4 volts RMS, or 95.4 volt peaks of the waveform, which equates to delivering 568 watts at 8 ohms, 758 watts @ 6 ohms, 1137 watts at 4 ohms. Tall order, but there are off the shelf modules available that an OEM can put into a chassis and satisfy requirements, depending on the duration, impedance curve, and content distribution. The loudspeakers might even survive it.

The fact that RP22 specifies ‘long-term’ output and levels 3-6 dB more, let that simmer for a while, is slightly concerning considering what it would take to get there. What kind of equipment could hold up? Distortion and compression levels are also one of those things we’re keeping on the shelf until later, because we’re still in the land of simplistic, but we should be aware that these limits do exist.

If you have these speakers, or something similar, and claim there is no distortion at these levels, even with your AVR offering far less power capabilities, I’m not going to argue with you. However, that claim would be very unrepresentative of a typical equivalent home audio loudspeaker with these specifications, and the math just doesn’t work.

Back that up to 4 meters, just because.

We’re now looking for 89.9 volts RMS for ‘peak burst’, 127.2 maximum available peak waveform voltage, correlating to 568 watts at 8 ohms, 1348 watts at 6 ohms, and 2021 watts into 4 ohms. (In the calculator, it rounds, so it’s not exactly half in presentation). Definitely not AVR power and outside of the range of most home audio outboard power amplifiers, by significant margin. You need either very serious esoteric home audio amplifiers, or very serious Pro Audio sound reinforcement amplifiers, not to mention the shameless flirting with loudspeaker damage.

Framework 5: The Boost Penalty

The Boost Penalty is the reduction of available system headroom caused by positive EQ gain.

Every decibel of boost must be paid for with additional amplifier output, additional driver excursion, additional thermal stress, or all three. EQ cannot create output. It only reallocates available output.

Let’s just make it a horror movie, and start introducing the blurry conveniently undisclosed inconveniences to our simplistic calculator, and see how bad this can get, invoking RP22 gremlins, responsibly disclosed by the founders of that document, Playback Contours and ‘Room Correction’ or any other applied EQ with positive ‘boost’ invocations. This is just an entertaining worst case scenario that would never happen, because people never want a response that deviates from flat, and room correction algorithms never boost. This example runs into Framework 6

Watch my horror movie cheek pulsate on one side of my horror movie face, as the horror movie tongue tries to break through.

Let’s do number 2 first. Enter the Gremlin, “Room Correction”, and the horror movie catalyst snack after midnight, lousy acoustics coupled with ignorance on the topic of best integration methods.

In this riveting sequel to “I brought home the love child of an MIT professor whose real genius was marketing cheap crap and selling it at a premium”, our hero finds some arguably very good speakers that sound nice with his favorite music in the show room, deservedly got great reviews, buys a receiver with the latest ‘Room Correction’ software that the salesman promised will ‘fix’ his typically suboptimal living room acoustics, plugs in the microphone and faithfully follows directions on the screen, and calls it ‘Good’. And in a lot of ways, maybe it is.

Joe75XP pointed out that 95% of room correction will be cuts. That may be true. I hope it is. I cannot confirm or deny in general, but it makes sense in terms of magnitude of adjustment, and most correction results I have witnessed cut with more amplitude than they boost.

In my opinion, it certainly SHOULD be true in terms of application of best methods. While you certainly CAN effectively and efficiently compensate for the magnitude peaks associated with standing waves by cutting these frequencies, and the audible benefit is pretty clear, boosting to try to overcome acoustic cancellations is a fool’s errand. It is both inefficient, and to me, it sounds weird.

Framework 6: The Black Hole Fallacy

The Black Hole Fallacy is the belief that acoustic cancellations can be solved by continuously adding more power.

Nulls are often geometric problems. Treating them as output problems simply dumps additional energy into a cancellation mechanism. You cannot fill a hole with watts.

I will not assert what anybody else’s room correction software will do in their room, but I can show you evidence of a one such real life horror story in which I participated, and it is not unique, and it was done by a pretty dominant player in the ‘room correction’ market.

Room correction eq results — Room Correction Boost and Cut Results

These are the results of front left/right channels when the surround sound processor decided to ‘correct’ what was largely the cancellation effects of putting loudspeakers in the empty cavities of an entertainment center, as well as the comparison of what it deemed an adequate correction AFTER that particular acoustic issue with addressed with absorption.

How much boost room correction can apply — SPL Boost Applied

I didn’t get any explicit data print out, but if I eyeball that peak ‘correction’ at several hundred Hertz, that looks to be about 9 dB, of a factor of about 8x the power output. Copying and pasting the peak next to the scale, it looks to be slightly more. But let’s run with 9 dB, and throw the required boost into our simple calculator for our 4 meter nightmare.

How much power does eq boost require — Power Calculations with boosted frequency and distance applied

We now potentially peak our left/right channel maximum power requirements over 8 kW per channel at 8 ohm, over 10 kW at 6 ohms, over 16 kW at 4 ohms. We haven’t even implemented a ‘house curve’, Gremlin #1 but we will, because we like it, and add some “Let’s just say,” to complete the ridiculousness of it all, that the overlapping playback curve boost aligns with the maximum ‘correction’ boost at a point where that playback curve contributes a mere 3 dB of EQ boost to the mix at that overlapping frequency, requiring a total of 12 dB of maximum EQ boost that keeps perceived levels in the dialog range about the same. Double all of the above. 16 kW at 8 ohms, 32 kW @ 4 ohms.

Not. Going. To. Happen.

It kind of seems like an impossible situation. Is anybody saying to themselves, “That can’t be right?” Well, that’s the math. I’ve shown my work. Show me an error and I’ll correct it.

So, Why don’t we see disaster all around?

So then, how is this disaster not actually manifesting around the globe, and those people buying the best-selling variants at Best Buy or something similar partnered with a brand spanking new AVR not coming back in droves carrying torches and wearing appropriated shirts of ‘not so much’ Aloha?

Well, for one thing, this example is admittedly absurd, to make a couple good points.

The first good point, made by Tom Vodhanel originally, is that actually guaranteeing that a system can ‘cleanly’ meet the maximum possible output requirements of ‘Reference Level’ is not a casual undertaking, and not a given outcome simply by buying ‘good’ products.
The second good point is that setup and integration methods matter a LOT, and might influence what gear is appropriate, and what is not.
- Speaker selection: If you’re going to sit relatively far away from the front speakers, and you want the option to play on the louder side of the ‘reference’ field, you might want something leaning on the higher sensitivity side, even those resembling or borrowing from ‘PA Style’ speakers, but maybe holding onto at least a vestigial priority for fidelity. Keep in mind, higher sensitivity isn’t free, or everybody would be doing it. What you get in higher sensitivity is paid for, one way or another, if not in cost, then in sacrificing other elements of sound reproduction. Consider sensitivity, but don’t fixate on it.
- Distance: Direct sound SPL reproduction at distance is exponentially expensive, both in terms of amplifier output required, but also in allocation of resources to deliver it maintaining the same level of sound quality. Consider the possibility that, if the number of listeners you want to accommodate doesn’t require a longer minimum distance, maybe you want an arrangement that allows you to sit closer, so long as it’s not TOO close. Yeah, it’s a juggling act. Welcome to the world!
- WATCH YOUR BOOSTS, even if you didn’t mean it. Even if the computer did it for you, EQ boosts SOAK UP headroom, or rather lower your overall playback ceiling, which means you’ll more likely crush that skull. Even ‘cut only’ boosts (get the curve by cutting EQ, but then bump it up with gain) do the same thing. Automatic room correction is potentially dangerous, if unmonitored. Because if it decides that the remedy for acoustic cancellation is to dump more power into a black hole, you’re setting the achilles heel of playback, at the systems weakest point. This is something human sound engineers have known NOT to do for many, many decades. That it’s showing up in modern software products made by software engineers who are supposedly experts is, frankly, pathetic.
- What can you do about it? Verify your results. Maybe your automatic room ‘correction’ algorithms are ‘Little Angels’, and haven’t eaten hamburgers or donuts after midnight. Maybe yours is still fuzzy and cute, and exclusively helpful. BUT YOU MIGHT WANT TO ACTUALLY CHECK. It doesn’t hurt to know, does it? If something was wrong, would you rather not know?

Why is the curtain still hanging behind the Wizard?

There is a lot going wrong out there, but lots of people buying ‘regular’ not so esoteric gear, running with automatic setups, and are happy with it. How can that be?

Let’s reconcile that with complicating factors that become pretty slippery in terms of specifics, because even though a lot of the status quo is queuing people up for a figurative train wreck, lots of people are fine with the results.

Potential contributing factors include;

‘Reference Level’ provides a known dynamic range platform where a director or mixer may utilize up to 105 dB in main channels, and 115 dB in the LFE channel. That does not mean that playback of any given content that utilizes the tools of ‘Reference Level’ will actually require these maximum values. In terms of hitting the limits of dynamic range playback capabilities, these are worst case scenarios that the audio content could touch, as a maximum, throughout the entire work, not typical or regular.
If you’re not watching a movie with ‘excitement-oriented’ sound effects, getting anywhere near 105 dB in any channel, or even collectively, might be extremely inappropriate. Even with action movies, those are ceiling levels, guaranteed not to be exceeded, not guaranteed to hit.
Many people consider ‘Reference Level’ way too loud, both in commercial theaters, and at home, and while they take some comfort in that they bought equipment that can play at ‘reference level’ (even if it cannot), they don’t actually do that. Why?
Subpoint possibilities:Possibility: Distorted sounds ‘Loud’.

Framework 7: The Distortion Trap

The Distortion Trap occurs when listeners mistake added distortion for increased detail, dynamics, or excitement.

Distortion is information added by the playback system rather than contained in the source material. Extra content “information” is not necessarily original information.

In that inspiring thread, RCohen pointed this out. Sometimes, particularly with ‘regular’ gear in ‘regular’ setups, when things start distorting, listeners don’t recognize it as distortion, they just hear it as ‘loud’. Sometimes they like it. It seems like extra ‘detail’, extra ‘dynamics’. It is, actually, extra. It’s just not original content. If a system only sounds ‘good’ when turned up, chances are that the listener prefers that distortion. If the listener doesn’t prefer that distortion, they turn it down, it sounds more comfortable, so they assume that louder means uncomfortably loud, because correlation must be causation, right?

NOPE! Case Study Example:

In this case, in a prior life as an audio equipment reviewer, I did a home audition review of a ‘THX certified‘ receiver, with ‘THX Ultra certified’ speakers, both certified, both proudly displaying licensed trademarks. I noticed that using the onboard amplifiers, calibrated and set to ‘Reference Level’, a scene in ‘Fight Club’ with a crashing airliner sounded uncomfortably ‘Loud’. It was objectively loud, in that you couldn’t easily have a conversation while it was playing. It also irritated me. If I turned it down to -10 dB from ‘Reference’, our more common playback level, everything was groovy. Smooth, easy, didn’t sound ‘loud’.

I might have erroneously concluded that ‘Reference Level’ meant uncomfortably loud with that content. If I had liked that extra ‘edge’, I might have erroneously concluded that until you play at ‘Reference Level’, you don’t get all the audio detail. But swapping over to a couple of outboard amplifiers that offered not just more rated power, but a metric $#! + Ton of more power supply, playing that scene at the same calibrated level, easy peazy, and it sounded pretty similar at -10 dB, or 0 dB master volume settings. Exciting, fun, but didn’t feel loud. It only became truly apparent how loud it actually was if you tried to talk over it. If I plug my numbers based on my best recollection of that scenario into that simplistic calculator as far as projected power requirements… It makes a lot of sense.

Amp power requirements to hit THX spec over distance — LCR Voltage/Watts Requirements vs Ohms

At -10 dB below ‘Reference’, that AVR was probably just cruising, a few licks above ‘Potato Power’. And to be fair to the AVR, it was a relatively good one, and -10 dB below ‘Reference’, while short of ‘The Reference’, is entirely usable. Everything sounded really good, and nobody asked to turn things up. Most non-enthusiasts, given control of the volume and setting it by ear and no indication where it ‘should’ go, would actually prefer that level over ‘Reference’. Here's the uncomfortable possibility: if you handed the remote to a hundred ordinary people and never told them where "Reference" was supposed to be, most would never find it. They would stop earlier. Happily. Then they'd watch the movie and get on with their lives.

It could probably have played a bit louder than -10 dB on the ‘Reference’ scale without significantly more audible distortion, but I did not test it to that level of specificity. If I were to make a SWAG based on specifications, optimistic rating tendencies of mass market AVR manufactures, and the fact that the loudspeakers were a 4 ohm ‘nominal’ load, I would guess it would be fine up to -6 dB below ‘Reference Level’.

What is disappointing is that the THX standard implied that proper selection within the ‘Badge Family’ would ensure that this wouldn’t happen. In my opinion a lot of good things resulted from the THX program, but a lot of the standards eroded as the ‘pay to play’ license requirements eroded on the verified performance side.

Possibility: If you don’t know, you don’t miss it.
Most of the people who buy this stuff are ‘regular’ people who don’t know what ‘reference level’ is, or that they’re ‘supposed’ to have it at that level to hear it ‘as the director intended’, and just aren’t interested in that higher range of SPL, and they’re better off for it, because they just play movies at lower levels, and everything just works fine. The dirty secret is that most people don't wake up in the morning dreaming of 105 dB peaks. They want dialogue they can understand, bass that doesn't sound bloated or anemic, and a system that doesn't punish their wallet, their spouse, or their neighbors. The rest of us—the enthusiasts, the reviewers, the standards committees, the forum obsessives—are arguing over the garnish while pretending it is the meal.

The practical performance downside of the logarithmic nature of dB and our hearing sensitivity is that it takes exponentially more output to get a little more perceived SPL. If you really want LOUD and CLEAN, both shouted, and truly so, actually coinciding and working in tandem, it gets expensive, quickly.

But the upside of that same relationship is that if you simply lower your SPL requirements a moderate amount, you can do what is otherwise pretty similar in sound quality to very expensive systems geared for high SPL at a fraction of the output requirements, and cost to produce it. 6 dB reduction means a 75% reduction of required SPL and associated loudspeaker and amplifier output. Within a specific system, it usually means significantly less distortion. Cheap sound quality upgrade for anybody interested. Free home trial too.

Possibility: Clean and balanced systems don’t need as much spl to convey excitement and that sense of ‘immersion’ in the experience.

Through my audiophile journey, when I discovered the realm of the legit ‘LOUD AND CLEAN’ sect of product offerings, high sensitivity combined with relatively high power handling, I fell in love with it. My friend Matt Potter, who I met in college as ‘That guy with the crazy stereo’, had a VW Golf and trunk space, introduced me to United Audio in Illinois, and to the Klipsch KG4 blew my mind. I put most of my summer job savings into a pair of Klipsch kg 5.2s, then quickly took advantage of United Audio’s full trade-in value upgrade plan and stepped up into Forte II’s, rated at 99 dB/2.83 volts. In my opinion, for what they were, they were pretty fantastic.

We had done some casual auditions in stores and friends' places with other systems, including Infinity Renaissance 90s (to appear again later), the best of the Infinity line that United Audio had to offer, at several times the price. We were convinced, high sensitivity combined with big boxes and bass reflex was where it was at, and Paul Klipsch was a god.

Those were a ton of fun. Had my life gone a different direction pursuing different interests, I might still have those to this day. If space and money were of no concern from then until now, I’d probably still have them.

But when I got out of college, and tried to get a ‘real job’, nobody seemed to be suitably impressed with my degree. So I took the job that was offered, selling audio equipment at an ‘Old School’ HiFi shop, “Mateo Hi-Fi”, the last shop to have an illuminated rotating sign. I think the owner, Fred Cohen, had been running it since before I was born. David Costa, the manager, saw some potential, and eventually even Jerry Berenstein warmed up to me and fed me apples.

I got a lot out of that job that I didn’t realize at the time, and I’m glad I got it. It provided many growth opportunities, and opened windows for professional development and advances. And I learned a couple things about audio that shifted my paradigm on playback.

The first was that audio equipment can have interactive effects and requirements.

We’ve covered the interactive requirements of amplifier power, loudspeaker sensitivity, and listening distance. That was relevant, in that you CAN get higher output with a lower sensitivity speaker, but it’s going to take some serious electronic chops behind it, not inflated rating novelty ‘power’.

But also, qualitatively, some speakers just ‘like’ some amplifiers, and some are fussier than others. A pair of speakers might sound pretty ‘meh’ with some amplifiers, but give it what it actually needs, combined with source content that can show off what it can do, and WHOA BABY, What is happening HERE?

A strength of the Klipsch KG 5. 2s and Forte II’s were that they really didn’t care about what drove them. A 45 watt Yamaha receiver, an 80 watt Yamaha receiver, a 30 watt NEC integrated amp, an NAD 105 watt power amplifier, a 225 watt Carver amplifier, maybe they sounded different, maybe they didn’t, but the speakers always sounded pretty much like what the speakers sounded like, regardless of the ceiling limits. Hey, that’s an aspect of good design, if you can get it. But, design is about balancing performance priorities. If you free yourself from high sensitivity, easy impedance requirements, other opportunities open.

Remember those Infinity Renaissance 90’s we had auditioned and dismissed at United Audio while I was in college? Mateo HiFi had those too, and David Costa, the manager, sold a lot of them. But he was really particular about how he powered them for auditions, because these things were fussy. They would make the power amplifiers sound different, even though ”That shouldn’t happen. ”

With the Carver amp rated at 350 watts into 4 ohms, which I had been using with my Klipsch Forte II’s, bleh, no matter the content, no matter how far from ‘clipping’ the output meters promised.

With a Hafler 9500 ‘Transnova’ power amplifier, same power rating, heck of a lot more power supply, those things grew some cajones in the bottom, but the midrange and top end sounded too ‘clinical’ for me to enjoy them.

Cary Audio SLM-100 ‘push-pull’ tube monoblocks, sounded gorgeous for vocals and decent with some bass that utilized heavy compression in the mix, but severely output limited and finicky with content. Some sounded great, some just ‘eh’.

Cary Audio 805 ‘Single-ended’ amps rendered vocals like nothing else, probably not accurately, but OH so sweet, open, and dimensional. Totally out of my price range, and bass was nice, but no oomph when required.

Onkyo M504 ‘dual mono’ power amplifiers, plinky and weak in the bass. Passively bi-amped to split the current delivery between the woofers and 200 Hz up array, darn decent. Still a little plinky, but things filled out a lot. Depth, dimensionality, warmth. Almost there…

And then Michael D’Or introduced me to his Aragon 8008bb. Holy crap. It wasn’t that I could play louder, it was that the speakers came ALIVE, all of them, all the musical content, to the limits of the recording, and you could HEAR the recording, mixing and processing techniques, microphone placement. It was kind of scary. The amplifier didn’t just drive the loudspeakers. When the amplifier drove the loudspeakers, and the loudspeakers reacted, the amplifier reacted back, and gave the loudspeakers what they needed. To put it in terms of a Cody Hiebert phrase, the amplifiers provided the reactance that the reactive load of the loudspeakers required, and it was freaking awesome.

Those Klipsch Forte II’s will always be special to me, and they will always be able to play louder than anything I will likely have in my home, ever again, and they can do it on a fraction of the budget, both in loudspeakers, and the electronics required to do it well, but once I discovered the rest of what HiFi really had to offer, I completely lost interest in the ‘shouting match’. I didn’t need to turn it up to hear more, because I could, with far lower levels, hear far, far more of the recording. The system was no longer trying to play over itself.

Which is the point regarding ‘reference level’ and how LOUD you may ‘need’. If you actually have a clean, quiet system at moderate levels, where all things are operating properly and in balance, the flip side advantage of a thoroughly clean and well-integrated BALANCED system, is that you don’t NEED to turn it up to reference level to hear everything clearly, and with things like ‘Dynamic EQ’ or non-trademarked auto-loudness compensation curves, which do ‘boost’ low frequencies, but not near as much as the overall volume cut it offsets, or a well-tailored manually set playback curve, it allows the tonal ‘fullness’ that accompanies our perception sensitivity tonal shift at higher playback levels. So it is possible to get more total satisfaction with less SPL, even if we are falling short of ‘Reference’ playback levels. Is it exactly the same? Well, no. Nothing is exactly the same in human experience. Even the same thing a second time through will be different than the first. But it is far closer than many realize.

Weigh what the difference is worth to you. Depending on circumstances, especially cost constraints and room capabilities, the end result holistic performance capabilities could actually be better by working within lower ceiling requirements. Compare that to the costs of what is required to do MOAR, maintaining the same sound quality in other aspects. If it’s worth it, go for it. Just understand that quality and the experience of immersion goes FAR beyond peak SPL capabilities, and make your decisions accordingly.

Real-World Complexity: Bass Management and Calibration

Shifting deep bass to subwoofers protects your main speakers, but stacked multi-channel tracks can sum together and push subwoofers to extreme, unintended limits.

Bass Management yields real benefits. While it doesn’t remove the Achilles Tendon from the Achilles’ Heel, it does shift a lot of that load, in terms of actual content, to the hammies, quads, and buttocks, muscle metaphors for the subwoofer(s). The higher the crossover point, the more ‘hot’ you run that sub, the more of the output load shifts away from the ‘main’ speakers and the AVR or outboard amplifiers supplying power to those main speakers.

How much? Some. Significantly some. The specifics depend on the contents dynamic distribution spectrum, and the playback curve. I think a general rule would be a mistake to rely upon as any kind of guarantee, because as Cody Hiebert would say, “Only content is content, and it’s all different, and all rooms are different”. So a generalization is no basis on which to establish anything resembling a performance guarantee.

I do think a lot of people presume incorrectly that bass management removes output requirements from main channels more than it actually does, more consistently than it possibly can. For example, in my living room system, I have 8” woofers operating from 80-320 Hz, and they give out WAY before my 18” subwoofers even break an easy sweat. The bass becomes a little muddy, the aluminum cones attached to the voice coil aren’t just warm to the touch, but radiate heat, and at one time, you could smell them. But if they had been trying to do what the subwoofers were doing, they’d have been toast LONG before 80-320 Hz pushed them to limits, and they never actually could anyway. So, benefit, significant, but difficult to quantify. But, whatever the specific benefit may be, bass management helps, most of the time.

In the context of contributing factors that explains the discrepancy between our Horror Sequel and what we see around us in real life, that’s all we need, significant help, most of the time. USUALLY, when you have an extremely loud sound effect, it is dominated by bass content. But not always. Hence, why you cannot rely on it to guarantee main channel playback capabilities.

A relatively uncompressed ‘CRACK’ or ‘SLAP’ may not benefit much from bass management at all, but such sounds often are not pegging the top of the dynamic range ceiling either, because most of the time, they’re severely compressed. Gunshots, for example, in real life, can be EXTREMELY loud, even without the low frequency concussion components of the ‘big loads’, to the extent that almost every audio system deployed doesn’t have a chance at accurately reproducing on-screen gunfire, and that’s absolutely fine, because WE DON’T WANT THAT. THE REAL THING REQUIRES HEARING PROTECTION.

On the other hand, slamming a door, particularly in a sealed room, certainly benefits from having the bottom diverted away from main channels. Something like the End Of Tomorrow opening bandwidth-limited square wave novelty system killer content, heck yes. So it doesn’t mean that your main speakers are given a reliable guarantee of avoiding heavy lifting, but they don’t have to do all of it, and they have a fighting chance of coming through a situation that otherwise might have overwhelmed them horribly.

Whiplash moment back from optimism: Any savings to the main channels provided by bass management are paid by the subwoofer in real costs. To illustrate this, our simple calculator in the range where subwoofers operate exclusively just don’t work very well, because;

‘Sensitivity’ often cannot be represented by a single number, as the output of the driver typically, without electronic compensation, is very far from ‘flat’. This isn’t a negative performance issue, as we often have complimentary room gain, and EQ’ing to tailor that relationship and playback curve is simply accepted best practice, so long as you’re not trying to overcome cancellations with it. But it is a complication that precludes simple calculation, and we cannot assume that ‘room gain’ of any given room offsets the ‘natural’ response fall off curve of any given driver/enclosure in equal proportion, because driver/enclosure alignment responses are all different. So, unless you actually KNOW the room behavior with particular subwoofer location(s), you can’t calculate output requirements with any specificity. See next item…
Room gain, cabin gain, whatever, what is yours? It’s probably not mine. Do you even know? You can calculate it, sure. Better software yields better guesses, but it’s still a guess. Have you verified that calculation with real measurement results? If you haven’t, I maintain the position that you don’t know.
Distance is no longer related to inverse square law, at least not reliably so, when you’re operating within a contained environment at very low frequencies. It’s complex and it’s different depending on the room and location variables. Good luck. I measured my room’s ‘transfer’ function with four optimally located subwoofers. It differs from inverse square law, a lot. That ‘Cabin Gain’ calculation trick you learned how to do in WINISD based on the longest room dimension? Nope.

What can be simply calculated, on a horror show level, worst case scenario, is that comparative additional output that subwoofer needs to be able to provide over the dedicated LFE channel maximum requirement of 115 dB, before EQ or running the subwoofer ‘hot’ as an ad hoc tone control, in worst case scenarios.

I’ve verified in mixers that when you sum signals, it is akin to summing voltage. That makes sense, as that is akin to acoustic, in-phase, time-aligned summation. It’s also how a Linkwitz-Riley constant voltage crossover works. It works well, and as promised, I might add, if properly applied with actual time-alignment, to be distinguished from simply adjusting delays until cancellations and peaks smooth out. This is how I would want it to work, by default, to optimize playback fidelity so that I’m getting bass levels I ‘should’ when combining channel content. Lucky me.

In all cases, electronic or acoustic, summing equal output of two equal parts yields a net gain of 6 dB. Total acoustic output of two sources of equal signal and identical level, if not physically co-located, may actually be 3 dB, in terms of total power output, but at positions where these wavefronts align, and the contributions are equal at those positions, it will be 6 dB. In our reference seating position, IF we do our setup correctly, (AND WE WILL) all channels WILL be truly time-aligned, and in phase, so 6 dB summation of low frequency left and right channel bass output, would be the target result, and this doesn’t change when you route all of that output to a subwoofer, or for that matter multiple subwoofers.

This isn’t magic, or some mathematical contradiction, because the total power output is in all directions, which reflects everywhere, and includes non-aligned output, which may be significantly below 3 dB, or even severely cancelling in some directions and locations, depending on the setup and the room. We just want those directions and locations to not include where people listen.

I mention this to explain the calculation that then projects potential output requirement ceilings of a subwoofer system when tasked by bass management to reproduce content not just of the LFE, but ‘main’ speakers on top of that. Worst case scenario: If some mixing maniac actually decided to get spicy and run full scale (0 dBFSD) bass to all channels in the system simultaneously. “That shouldn’t happen. ” As much as that is a saying I like to mock, as I have repeatedly seen what “shouldn’t happen” eat that pizza after midnight, in this case I agree with it. But, I also cannot guarantee that it will not. So worst case scenarios, though admittedly highly unlikely in terms of worst of the worst…

Work shown so you can catch me and I can correct it.

How signal summing affects SPL — Summed SPL levels

What happens to that 115 dB theoretical protective ceiling on subwoofer output when it was ‘only’ responsible for handling the LFE? Well, if we’re just adding bass from 2 channels, not a big deal, as both channels were 10 dB lower individually so; 115 dB maximum theoretical requirement becomes… 119.26. That’s 4.26 dB more. Significant, but not implausible, but should be factored in, if we’re in the position to do the factoring.

Bass from 5 main channels simultaneously routed in our “This will never happen Scenario”? 123.24 dB. Whoa. It adds up. Good thing that doesn’t happen. What kind of doofus puts maximum bass into surround channels. That would be really dumb. But dumb doesn’t mean impossible. We cannot guarantee dumb never enters the building. Say 7 channels out of a 7.1 format get loaded up, 125.14 dB. Huh. Probably won’t happen on content, but it could, theoretically. So, with 7.1 formats, we know the potential ceiling on output may be beyond the commonly cited 115 dB of the LFE channel alone. How much? Somewhere between 0 and up to 10.14 additional dB with 7.1 content. More real full-range channels means more potential subwoofer load.

Does uncertainty bother you? Well, if you can guarantee pristine playback through the operating range up to 125.14 dB, no sweat, you’re prepared for whatever 7.1 may offer. Do you like a nice lift on top of ‘flat’ playback? ‘Reference Level’ assumes ‘flat’ playback.

If for instance you like the product of Harman’s subjective playback research, the ‘Harman Curve’, depending on which version of that you prefer, you might be adding 4-8 dB of additional bass level, globally, not just to your subs, but primarily to your subs, in terms of content distribution and impact.

For any interested, this article summarizes the origin of the Harman Curve, in the context of earphone playback.

Because the music industry in particular is so NOT standardized, there IS no actual playback standard, which explains why so many people dislike ‘flat’ playback. The Harman Curve was established by testing curve options and determining which curve, for most people, for most content, was determined most enjoyable. It is not a matter of correctness, it is a matter of availability and what is largely compatible with many listeners.

Harmen target curve — Harmen Target Curve

If you like a more moderate ‘Harman Curve’, or one of the many available proprietary manufacturer-supplied playback curve maybe that’s 4-8 dB more.

The Dirac website has a bunch of available target curves, each with their own rationale. I would repeat, none of them are inherently correct. It is simply a matter of availability and acceptance. If you like something better, rock on. Resource: Dirac Target Curve

Again… huh… with that 4-8 dB lift, the suggested bass lift range by Harman, a mainstream and reputable player in the audio space, that 125 dB worst case scenario now ranges from 129-134 dB maximum.

There are other versions of target curves, all with their rationales, which we need not dissect, but the key take away as it relates to output levels, is that they all ask for more bass, and they’re all different, and if you want to actually design a system that can handle your target curve or preference at a particular targeted output level, you’ll want to know what that is, and incorporate that into your calculations, unless you’re okay with ‘try and see’.

I like ‘Try and see’ too. It is in fact the most reliable method, but it takes work, and utilizing it means iterations. So, if you’re going to shift your planning into more of the ‘Try and See’ realm, allocate time and budget for those iterations.

Again, these worst case calculations are not likely. Bass is usually routed to the front speakers, because in commercial cinemas they are the only ‘main’ speakers large enough to handle bass, helped by the LFE speakers, usually producing redundant bass. But stuff happens. If we took the 4-8 dB Harman curve lift and applied it to Left/Center/Front main channels going full tilt in tandem with the LFE, that would sum out to 120.8 dB, meaning that with an application of a Harman curve, that would bump requirements to 125-129 dB, in round numbers.

We can see where with the application of playback curves ‘headroom’ gets turned into occupied space, and then potentially a concrete ceiling falling down quickly. How fast, how hard, depends on the content demand. This is where brutal tests like the Interstellar wormhole scene really tend to show the limits of things in not so nice ways.

Checking back into a more ‘comfortable’ reality: possible mitigating factors as to why bass management hasn’t turned all ‘average’ subwoofers into a pulpy mess with completely unacceptable performance results.

The worst case scenario is extremely unlikely. We cannot guarantee that it won’t happen, but we can expect that most times, it won’t, and the ‘perfect storm’ is usually just a wet blustery downpour compared to how bad it could be, hypothetically. A leaky roof that maybe nobody even notices, or you can just leave a pot in a spot and forget about it. That is very different from the roof divorcing your house to ‘find itself’ in the wind.

Lots of subwoofers have ‘limiters’ of sorts built in. Some have electronic limiters or more sophisticated protective circuits, the best of which maximize the usable linear range but prevent damage, hopefully subtly. Sometimes the limiters are mechanically built into the drivers, not technically ‘limiters’ in the strict definition, but with progressive suspensions that bottom out gracefully without exploding will both save subwoofers, voice coils that give up their effort by backing off available motor force at high excursion, padding that acts as a bumper to deliver a ‘burp’ instead of a WHACK at mechanical limits…

All of them ARE distorting and compressing, but it may not be obvious, and the clever ones stay out of the way as much as possible, and listeners may not even notice. Unlike a tree, if a subwoofer falls down, momentarily anyway, there is nothing left on the ground to discern later. The moment has passed. Human beings are pretty tolerant to transient distortions. We don’t have the processing bandwidth to notice something is ‘wrong’ if only exposed for a short sample, until things get really bad.

But returning to the point of ‘short-term distortion forgiveness’ in humans, that is why those CEA2010 test ‘Burst’ numbers have distortion thresholds that, if you actually listen to them with a decent pair of headphones, at a constant continuous sine fundamental with REW adding the harmonic content to replicate those distortion levels, would sound pretty freaking atrocious. We’re going to get to that….

If you’ve hung out this long, I thank you for your patience, and hold onto it, because we’re going All The Way.

That begs a question shelved until later, are we there yet?

As Mitch Hedberg once said, “Coming Soon…”

Distortion related to subwoofers tends to be in ranges less sensitive to human hearing. It’s not that you cannot hear these distortion products, but that they may be far less objectionable, less ‘harsh’ or ‘strident’. It may come across as ‘character’. You may even mistake the distortion products as part of the sound track. Without an actual clean reference of comparison, you may even mistake moderate distortion products of a pure sine test tone as the pure tone itself, or distortion of content as extra ‘detail’.

Heck, some people LIKE some of these types of distortion. Jeff Mery, of Bison HTA, once did a demo for a client comparing two subwoofers I know intimately, and the client preferred the unit I knew and verified to be technically inferior in all ways I could discern. To be fair, it was mass-produced, made in china, delivered for $400 shipped, and a fraction of the selling price of the better quantified unit. That particular listener preferred the ‘character’ of the unit with distinctively interesting compression/temporal distortion. Hey, his preference isn’t wrong, and if that means he can spend less to get what he wants, cool. I’m not going to try to convince him otherwise.

Such is the nature of human hearing, especially if we’re talking about scenes with things blowing up, and the distortion products may resemble a lot of the content to the brain, and that distortion is subsequently partially ‘masked’. It is not that it is actually inaudible, but if a listener doesn’t have an undistorted reference, or know what that type of distortion ‘sounds’ like, that distortion will remain unrecognized as such.

In the case of a distorting power amplifier ‘clipping’ while driving a subwoofer, you get an added bonus, in that the woofer’s bandwidth limitations will partially filter the amplifier’s higher frequency, more objectionable, distortion components.

Unlike a passive full-range speaker, where the passive crossover simply routes the amplifier’s clipping harmonic products, up to the midrange/tweeter as appropriate, where they can reproduced with full loudspeaker efficiency, the subwoofer driver ‘tries’ to reproduce everything coming out of that dedicated amplifier. The upper end of those distortion products are often beyond the subwoofer driver’s bandwidth, so the subwoofer driver effectively lowers the audibility of those particular amplifier clipping distortion components. That’s not to say that the amplifier distortion is inaudible, but rather less objectionable.

So back to the arm-chair of practical comfort. It’s not really a horror movie. Play stuff, enjoy, it’s fine. Unless it’s not, but you get to decide.

We can see how there are mitigating factors that make the “Chasing The Reference with ‘Normal’ Gear” not necessarily a nightmare, but possibly just suboptimal, still very workable, one way or another. Without making very careful informed choices, fidelity is not guaranteed, but output is likely ‘usable’ for most, especially if we let reference level have some distance ahead of us and just back down a bit.

And how much of a ‘Reference’ should ‘Reference Level’ serve as THE reference anyway?

As the article David shared from Residential Systems mentions, and I’ve noticed myself, the standardization of playback levels on streaming services, between the service, the apps running on a device, and the device itself, seem ALL OVER the place, so much as to make ‘Reference Level’ as a reference useless with this content, which has become, most of what we’re watching. It really becomes, “adjust it until it suits you, and then see where that lands,” which is in fact, no reference at all. Hey, I used to do that. If it worked for VHS tape, you can probably get it to work with NetFlix.

The industry spent decades building a cathedral around Reference Level, only to hand the keys to streaming platforms that routinely bulldoze consistency the moment content leaves the studio. If playback level is no longer reliably standardized from service to service, device to device, app to app, then much of the conversation collapses into a beautifully engineered theory with a broken supply chain.

https://www . residentialsystems.com/features/home-theater/rethinking-reference-level-in-residential-audio

I’m sorry to say that I’m not coming to any definitive conclusions, other than “be aware” and “in depth understanding and involvement, and independence, can pay off if wielded carefully. ”

Oh, which kind of brings me back to the ‘Performance Level’ RP22 tier standards, advocating 'head room’.

On the surface, ‘headroom’, in the pure sense of the word, is only safety margin, it doesn’t actually get you anything if you don’t use it. So, planning to have some of it is a prudent safeguard. If you actually have cost constraints, at all, one of the best values in audio is not paying for what you don’t need.

But margin, particularly when determining what you need with unreliable information, can be a good thing. We can see how extenuating circumstances, or simply things not accounted for, or missed by poor accounting, can soak up that safety margin of headroom, so that it is no longer headroom. And, the data which we rely upon to get those direct sound SPL predictions is not always exactly correct.

Further complications to safe calculations:

Tom Vodhanel pointed out that often, loudspeaker specifications are not reliable. Maybe the sensitivity is stated in a way that is misleading, or just a lie. Maybe it states a ‘nominal’ impedance that does not represent the most difficult lower impedance presented to the amplifier, and while that amplifier may be able to meet rated power at 8 ohms, at 2 ohms it completely falls apart prematurely relative to the voltage output which is the sensitivity reference.

I’ve measured ‘Class A’ rated ‘super speakers’ that had impressive 95 dB/ ‘1 watt at 8 ohm) sensitivity, which were nominally claimed an ‘easy 8 ohm load’, that in a particular setting on the woofer’s electromagnet, actually dipped below 2 ohms in the middle of the bass region.

Graph of Impedance changes over Bandwidth

That ‘8 ohm easy load’, in the 20-500 Hz range, typically the most demanding part of the audio content spectrum, is more like a demanding 4 ohm load, that drops to 1.9 ohms.

If we normalized equivalent ‘1 watt’ sensitivity to a nominal 4 ohm voltage, that would be 92 dB per 2.0 volts, 1m, or 1 watt at 4 ohms. Not bad, but half as impressive.

If we look at the achilles heel of the amplifier power delivery and the loudspeaker’s thermal handling capacity, at 2 ohms, 1.47 volts, equivalent to 1 watt a 2 ohms, that would kick them down to 89 dB. Can the amplifier deliver 4 times the power at 2 ohms as the rated 8 ohm delivery? If not, the ‘nominal’ SPL estimate falls apart. The numbers tell a nice story which is ‘nominally’ true, but not the entire story. Interestingly, I did not see these real impedance plots in the sales literature.

It is also true that many amplifier specifications are not reliable, or not actually indicative. As those who peruse Audio Science Review may know, many power amplifiers do not actually meet rated power specifications. Some don’t make them under any conditions. Some just don’t make them under all conditions. It may meet power ratings at 1 kHz, but 20 Hz, or lower, it just falls apart. It may be able to swing 90 volts into 8 ohms, but into 2 ohms, the rail sags and it clips prematurely. And about that power supply voltage rail, if it’s not stable, it might suffer what James D. Johnston, PhD, once described as ‘gain shift’, a potential primary source of audible shortcomings in power amplifiers, which Cody Hiebert (Harbottle Audio CEO) refers to as power amplifier compression, which Stephen Mantz of Zed Audio confirmed as an issue related to long duty cycles and inadequate rail stability, that nobody in the Standard Industry wants to acknowledge, let alone disclose.

All of these predictions assume linear operation of all devices, i. e. , no compression, and all components working as they ‘should’, i. e. , effectively near perfect. We do not live in a perfect world.

So, in an imperfect world, maybe that headroom does get you some kind of quality improvement, by keeping things operating more safely within the bounds of what it ‘should’ be able to do, if we did live in a perfect world. If it can go to 100, and we stay below 50, that bottom 50 should be close enough to perfect, right? Do you buy that?

But wait, there’s more! Not only do you get a complimentary Robo Vegematic worth buuko bucks with the next 100 orders it’s reasonable to assume that everything else being equal, things will have less distortion the farther you are below a system’s playback limit.

So maybe RP22 guideline suggest more ‘headroom’ with higher tier performance standards in part because it may imply better quality as a result before those maximum playback limits, as a result of shifting up maximum SPL limits beyond required playback levels, even if that headroom is in fact, from an SPL standpoint, “wasted”. It’s a very reasonable speculation. But I think it’s also dangerously unreliable. I briefly touched on this earlier. Now we’re going to hammer it in.

Manufacturing Perspective (Cody Hiebert). RP22 is incomplete, it says in the documentation they they have not decided on a measurement protocol for fullrange or subwoofer speakers. As such, if I was a CI, I would not start referencing propositions as standards until they are actually proven. RP1 is not available to manufacturers as a completed work, and RP32 is being deployed as per some rumblings out there in the ether regarding training and workshops. So my question is, how is the in room work being designed and verified if the manufacturers aren't able to verify targets because procedures aren’t locked down? There seems to be a gap in how CEDIA is operating that's a little too big to ignore.

Specification Reliability: Deconstructing CEA-2010 Standards

Subwoofer burst tests allow extremely high distortion limits, making short-term lab scores a terrible predictor of clean, long-term performance. Two subwoofers with identical test charts can sound vastly different during a continuous movie scene.

The issue I'm raising is this, and "Soon" is now.

About that reasonable speculation, about MOAR meaning better, because distortion decreases as you fall below ‘limits’, however you define them.

If the ‘Clean’ output limits of different tiers are all above actual playback requirements, and a higher tier provides better audio performance (assuming all else is equal), because as you move farther below the ‘clean’ output limits, distortion decreases, this acknowledges that ‘Clean burst’ distortion limits that may or may not be audible in transient conditions, are not inaudible with more continuous ‘average’ program levels.

I am picking out an anonymous subject from the data-bass files, because I’m not looking to pick a fight or hurt any butts, but I think the point is salient enough to illustrate with a real example… This is real data. This is a real product. This was tested by an impartial, third party, expert.

A unit of a highly regarded subwoofer, measured in 2016 that used dual 18” woofers, the CEA2010 maximum output limit at 63 Hz was 126. 4 dB, 2 meters ground plane. That’s a lot for a single subwoofer. I assume they use 63 Hz to work around test equipment with poor grounding schemes. THD measured 10.6%

CEA2010 burst test results — CEA2010 subwoofer burst results

Cool, nice. CEA2010 says that is the ‘clean limit’, or maybe it is the ‘practical limit’, which I saw in a document. The available data didn’t specify which harmonic, which matters a lot. But if we look at the 120 dB sweep that breaks out Harmonic distribution, it is dominated by the second harmonic. COOL! Much less audible than a third harmonic. It is also, as we would expect, lower in distortion dropping about 6 dB lower, down to 6. 38% total

THD distribution — Harmonic Distortion Distribution

Harmonic distortion 2nd order — 2nd Harmonic Distortion Distribution

Great. If CEA2010 limits are the limits for ‘clean’ or ‘practically clean’ short-term bursts, (argue amongst yourselves whether it is or isn’t if you wish), can we assume that below that any distortion is inaudible?

I think clearly not.

Subtopic: Example of the problem (or just one of several problems, this one being very easy to demonstrate directly) with CEA2010 used to infer sustained output sound quality performance.

CEA2010 is a short burst test that puts limits on the amount of non-signal (anything not signal is noise or distortion) allowed based on a table that allocates limits relative to the signal, expressed in -dB, of total energy in given bands, related by ratios to the fundamental frequency, from 16 hz spanning up to 10 kHz.

I paid for the CEA2010 document to get the information from the mouth of the horse. I don’t know that I’m allowed to redistribute their content in any way, but I can show my calculations based on Table 1 in the ANSI/CEA-2010-A document published July of 2012.If you’d like to verify my results, the source is available in the link above.

I determined limits and bandwidth based on that table, in this example, for a 60 Hz Fundamental. UP TO these limits, CEA-2010 says ‘usable’.

CEA/CTA 2010 Bandwidth Limits as per ANSI/CEA

CEA 2010 standards have often been used as a metric for subwoofer ‘Burst’ capability and ‘practical limits’, i. e. , distortion by at levels that a lot of people will still find usable. While we could argue how it could be better, and how well it applies to predicting that, I think that’s a reasonable position. I’m not going to open up a detailed critique of CEA 2010 testing procedure, and the ways it falls short as a metric for ‘burst’ or ‘practical’ peak limits related to either real content or audibility of distortion and noise components as it relates to the human auditory system. We can shelve that for now, and agree that it is a standard often used, and there is some sense to it used in that context, and a reasonable rationale behind it.

But you think it guarantees inaudible distortion at and below that point… No. Seriously, try it for yourself. Matt Potter was, after all, from Missouri, the “Show Me” state. Let me show you.

Seriously! NOW!

If you don’t have REW, get it, and donate something if anything you do with it provides you with some value. You can feel good about it, and it’s the right thing to do, even if it’s just a few dollars or quid, or squids, or whatever your local currency.If you want to know what Harmonic distortion sounds like, just add it. If you click the ‘signal generator’ Icon, you can enable the ‘Add Harmonic Distortion’ check box. I would suggest adding dither, as high in resolution as your playback allows, because if you don’t add dither, you get quantization distortion instead of quantization noise. Quantization noise is just a slight increase in the noise floor, and with 24 bits, your system playback noise floor is going to swamp that out anyway. Quantization distortion will add potentially audible distortion, and since we want as little distortion as possible to hear harmonic distortion, we’d rather not have that.

This tool allows you to adjust, real-time, the levels of each harmonic, and with each check box, allow you to hear the sine wave with and without harmonic distortion components

Click on the ‘Signal Generator’ tab, set the options as shown, go to town.

How to set up REW to hear harmonic distortion components

This is what generator settings would look like with all harmonics set at CEA2010 limits. This is what it allows. Because CEA 2010 groups 4th and 5th harmonics, 6th, 7th, and 8th harmonics, and 9th upwards all together, measured as Harmonic distortion + Noise within defined bandwidth limits, I arranged this twice, first with all energy put into the lowest harmonic of the respective groups (4th and 6th), which would be less audible, as lower harmonics are less audible than higher harmonics. If you’d like an example, you can slide the distribution weighting to the other side, entirely into the uppermost harmonic of the bandwidth group, the 5th and 8th harmonics. Just adjust the controls as such…

REW has Harmonic Distortion Controls and Levels — REW Distortion Controls and Levels

Note that in the screen shot of REW harmonic levels, the 7th harmonic, I did not set the value, because I wasn’t doing that test, but if you wanted to know what the seventh harmonic sound like at CEA 2010 limits, set that level to -35 dB like the 6th and 8th harmonics, enable the 7th harmonic, and disable the 6th and 8th.

If you do the test in REW, you can turn off and on harmonic distortion, real-time, to do pure versus distorted sound, as well as turn on and off harmonic contributions individually, as well as adjust relative levels. I would encourage playing with it. It’s really informative, and it clues you in how when some people claim to be able to ‘hear’ 10 Hz, maybe it’s not actually the ‘note’ they’re hearing, but the harmonics of that note, and if it’s a pure tone, it’s harmonic distortion.

You can also do this to determine your personal distortion sensitivity.

I did this test with the test signal set to -20 dB, volume ‘all up’ on my headphone port. Headphones were Sennheiser HD-280pros, and I’ve verified that output at 0 dBFSD at the headphone output is 2 volts, so a pure sine signal would theoretically be at about 0.2 volts on each headphone input. Use that information or discard it as you like.

The specific laptop used is NOT especially good for audio quality. If you’re doing any kind of scientific testing for determining reliable values, I would use a better audio interface device, as well as measure the performance of the system, and I would do double-blind testing with a large sample of subjects. You know, like laboratory-grade experimental methods. But in this case, the differences are so absurdly drastic it really doesn’t matter.

If you want to use your home audio system to play this back at levels that it’s distortion should be low enough to make this obvious, and you still have plenty of output to hear it, calibrate the tone to be presenting 2-2.83 volts at your loudspeaker terminals. That’d be equivalent to 1 watt at 4 ohms at 2 volts, and 1 watt at 8 ohms with 2.83 volts. It doesn’t need to be exact, but it will be loud enough to clearly hear, and any marginally competent system should have distortion levels low enough to make the difference painfully obvious. If you cannot hear the difference between a pure sine wave and a sine wave with harmonic distortion at CEA-2010 limits at a moderate playback level, something is wrong. Visit an audiologist. Maybe they can help you get more out of life.

Point being, for sustained output sound quality, CEA-2010 limits are entirely unsuitable if fidelity is of any kind of concern.

If you want, You can put these files together on a playlist and compare them back to back, already prepared. These are a Pure 60 Hz sine wave, and a sine wave with harmonics added at CEA-2010 limits,

Use our test signals to hear harmonic distortion components. — Play the distortion files through you audio player

Back on topic: Looking at our subwoofer comparison, where CEA-2010 performance is near identical.

So if CEA2010 Harmonic distortion limits are irrelevant for sustained output sound quality, from a fidelity perspective, let’s walk down the road a little more…

Having run an informal test and adjusting levels on REW and clicking add harmonics or not, adjusting them to levels were I’m confident I cannot hear a difference, I KNOW that regardless of harmonic spectrum, there is no way that 10.6% THD, even if it was entirely 2nd harmonic, 120 Hz, the LEAST audible possibility, is going to be inaudible on a comparatively sustained (in the realm of seconds, or even less) basis.

I did casual tinkering with REW, a pair of HD-280 Pro headphones, and my janky laptop audio output, at -20dB referenced to 2 volts into the headphones at 0 dBFSD on ‘full volume’ (100), but my windows output volume was set to 30/100.Information is disclosed for disclosure’s sake, but you don’t really need to do anything with it unless you wish to replicate it or offer constructive critique.

Based on this informal experiment, my personal threshold for 2nd harmonic on 60 Hz, limited by the masking effects of any distortion introduced by my headphones, as well as any damaged incurred by my auditory system over half a century of living, suggests that the point I am confident that I cannot hear the second harmonic of 60 hz at this specific situation is about 2%. 3rd harmonic 0.79%. It goes on down as we climb up the harmonic ladder, but this is just an example, not a personal buying decision. If you really want to be thorough, look at all harmonic data, all frequencies, all amplitudes, and generate your own corresponding distortion detection thresholds. Good luck, and have fun!

REW Distortion Controls set up for distortion limit comparison

As you move up in order, it takes less and less magnitude of distortion to hear it, and it sounds less and less like the original fundamental. My results conform to what auditory researchers have told me. Go figure. For ME, this is likely a worst case scenario, specifically for HARMONIC distortion (which does not include all audible distortion, for sure), where I am comfortable saying that I cannot tell a difference, the scenario where this kind of difference would be easiest to detect. At that threshold, with that setup, the HARMONIC distortion levels cited are by definition inaudible with a ‘pure’ sine wave tone. Maybe if I got some cleaner headphones or used a better audio interface, who knows, maybe that could shift lower, or maybe not. If so, probably not by much.

So to clear, these thresholds I’m using are just for example, just for 60 Hz, just for ME, at whatever output level HD-280Pro headphones output 60 Hz and reproduce harmonic frequencies above that with a 0.2 volt (-20dB to 2 volts) input, then reduced by windows from a value of 30 from 100 on the volume slider. I haven’t measured what that is. I don’t know if this is a good standard, but it is the standard I’m using for this illustration, and in this example it is clearly a far better standard than CEA2010 for sustained playback, even if arguably over-stringent.

Here are the results of my own listening test and a set of audio files from Harbottle. Run a decent pair of headphones and moderate (not loud) output to find out your own distortion tolerance levels.

To be fair, I don’t think CEA2010 was ever actually presented or intended to be used as the limits of audible distortion, let alone limits of audible distortion for sustained playback, and it was created by a smart guy who had good reasons for it being what it is. But I bring this up because it seems a lot of people THINK it represents something it doesn’t, the top end limit beneath which fidelity can be assumed, and it is CLEARLY NOT THAT. Let’s agree that my custom standard for this example is tight, and it’s a high bar, and that anybody is free to select another. I just want to show this trend in comparison, with real data, from real products. I needed a more stringent, relevant standard, to contrast real results as it relates to actual sound quality of sustained output, so I made this one.

Manufacturer Perspective (Cody Hiebert). Standards represent use case. If the standard does not represent how the product is used and the products of the device (audible output, in this case) then it is not a standard, it is something else and by its own virtue it is now pointless. So what was CEA2010 designed for?

As a manufacturer, the fact that CEA/CTA2010 removes compression data before the CEA distortion limit and does not show compression characteristics, when it is actually collected during the test, is an admission that compression based efficiency is a real issue in linear playback in burst.

So here we go, back to the measured data.

Well, as we lower output, distortion drops, as it does with most loudspeakers.

At 60 Hz (keeping apples in the apple basket), at the 110 dB nominal sweep, THD distortion drops to 2.58%. Not bad, but I can definitely hear the difference.

THD results of dual 18 inch subwoofer — Data-Bass THD result

Okay, note quite there yet, but potentially close.

Drop down another 5 dB in output to the 105 dB nominal sweep.

Hey, there we go! 1.57%, dominated by second harmonic, 105 dB, 2 meter ground plane, and the subwoofer response is flat, so it’s actually 105 dB in real output at 60 Hz. (dB nominal values are referencing 50 Hz output as per Data-Bass convention).

And we have harmonic distribution data on the distortion, cool beans! At 105 dB sweep, 2nd harmonic is only at 1.38%, well below our stringent sustained audible distortion limit.

Data-Bass Dual 18 subwoofer 2nd harmonic distortion

Checking the third harmonic…

Data-Bass Dual 18 subwoofer 3rd harmonic distortion

At 0.713, safely under, though not by as much margin. It’s hard to say how much higher this thing could play before the third harmonic crosses 0.79%. I did linear interpolation, which isn’t dependable, but it’s something. Based on linear interpolation between 105 and 120 Nominal dB sweeps, 3rd order distortion would be expected to cross that threshold right before 106 dB, but close enough that we can give it 106 dB, 2M ground plane.

Are the details of this actually important, to the general point? Well, yeah. This subwoofer, at 63 Hz, will play up to 126. 4 dB, 2m ground plane, with CEA2010 limit definitions, so it can burst like a BEAST with what many might consider ‘practically acceptable’ levels of distortion. But in terms of harmonic distortion levels I’m confident I cannot differentiate between some and none, that 60 Hz limit is only 106 dB. Is this good performance for a dual 18” subwoofer that boasts a 4000 watt amp, costs $3k in 2016 dollars?

Maybe it is. Maybe it depends on how I’m using it, and what I like. It was certainly highly regarded by many. That could make for an interesting separate discussion, variations on what people like, and how that sorts out with objective performance behavior, but that’s not where we’re headed….

Contrast comparison…

Subwoofer, single 21” driver sealed, CEA2010 limits, amplifier limited, 126 dB at 63 Hz, 2M GP. It cost about twice as much as the first subwoofer.

CEA2010 Burst result for 21 inch subwoofer

CEA2010 burst limit is pretty equivalent, 126dB, 2M ground plane, 8.1% THD. Defining capabilities based on CEA2010 burst capability, 0.4 dB less capable of ‘practically usable’ output, 126 dB vs. 126.4 dB maximum output limits, so maybe call it a tie in performance, and the first doing it for half as much, clearly the better value? You could say that. I wouldn’t.

Now this is where things get interesting.

If we search for TOCM (That’s me) limits at 60 Hz, laptop/280 masking protocol, to make it sound all scientific, the arguably over-stringent, but equally applied harmonic distortion limits adopted for the sake of illustrating “Soon”, transforming it into “now”…We want at least under 2% THD, at 60 Hz for the self-important TOCM standard at 60 Hz. Period! Done. 115dB nominal sweep THD is 1.59%

Data-Bass 21 inch subwoofer distortion results

That’s actually at 117. 4 dB at 60 Hz for the ‘115 dB’ nominal voltage sweep (Normalized at 50 Hz)

Data-Bass 21 inch subwoofer output results

But oh no! it’s dominated by the third harmonic, which is running at 1.53%. that’s about double TOCM Laptop-280 masked non-detection limit. Bummer. Keep bringing it down then.

Data-Bass 21 inch subwoofer harmonic distortion results — Data-Bass 21 inch subwoofer distortion results

Unfortunately, there is no distortion spectrum data below the 115 dB nominal sweep, so we have to go back to interpolating where we think it might be…

If I impose linear interpolation versus the 122 dB harmonic distribution data, distortion falls with output so quickly that predicted distortion meets TOCM masked 280lap protocol limits at 114 nominal dB, which we would interpolate to be 116.4 dB.

So, in short, essentially these subwoofers offered the same ‘Practical Use’ Peak Burst limits as per CEA 2010 standards at 63 Hz, but the sustained clean playback capabilities by TOCM 60 Hz harmonic distortion standard has more than a 10 dB difference in capabilities, for only twice the price, and with less occupied enclosure volume. In other words, you would need 2.5 of the dual 18 inch units to keep up with the single 21.

By one definition, at one frequency, a metric for peak burst capabilities, they’re nearly identical. By another definition, near the same frequency, sustained extremely clean output with inaudible harmonic distortion levels that I cannot detect under the most demanding condition, one is a freaking serious machine, and the other, not really in its element, so to speak.

Food for thought.

This is just one slice of a picture, one dimension, one standard. While it certainly is a ‘real truth’, it is also most certainly not the ‘whole truth’. I’m not naming names, because that invites arguments based on favorites and subscription service mentalities, and the names aren’t important. It is real data, and it illustrates a point, a point transitioning from “Soon” to “You’re reading this now. PLEASE LET THIS SINK IN. ” A point that touches on the ever-present reminder that reality is, indeed complex, and a single data point, taken to represent a whole picture, is potentially sometimes worse than none, BECAUSE THE LINES YOU DRAW FROM THAT SINGLE POINT CAN BE COMPLETELY WRONG.

Conclusion

The point is NOT that one of these subs is better than another. Shift your metric and priorities, can you can probably find something in one, either way, which you find superior. In the general public space, they were both highly regarded as top tier performers, and deservedly so. If you happened to own either, and you figured out which ones I’m referencing, and you like either of them, I wouldn’t suggest you change a thing.

The point is also not that my particular test is better than CEA2010, and that we should use it instead. It’s different, it applies to something different entirely. Different playback context, different standards, different, different, different…. I just invoked it because I had it, and it addressed something that the CEA2010 did not, something very relevant that CEA2010 totally misses, because it wasn’t designed for that. If you wanted to develop your own, collect data that varies with Frequency and SPL. If you want the shortcut so you don’t have to do all of this, copy the homework of somebody who did the work to define standards that align with your own, and solicit their advice, maybe buy their products if they offer them.

The real point is that depending on the particular standard of limits, either of which may be appropriate for the use context, one limit does not scale against the other in either a linear or consistent manner, and it’s NOT easily predictable.

As such, engineering systems for improved performance quality with a disproportionate emphasis on how much ‘headroom’ you have designed into the system, beyond what you’re actually going to use, while perhaps based in logical reasoning, is prone to failure. If the market misinterprets guidelines encouraging ‘headroom’ as a first priority, they unintentionally pressure manufacturers to sacrifice performance in ways which will degrade actual audio performance, qualitatively.

My point is now, “Now”. One data point does not make a curve, and curves are all different. Some of them are drastically different. Many of them are very important, if not to the end users, to how those performance metrics impact quality of sound reproduced.

Where the industry goes with this, I don’t know. But I see more work to be done, if we really want to serve the interests of the end user, as opposed simply manufacturers selling reassurance by numbers.

Key Points for Digestion:

Whether ‘Reference Level’, or RP22 recommended maximum output capabilities, one cannot assume that decent or even very good equipment will meet these output requirements. If meeting these requirements is important to you, at the very least, doing some basic math is warranted. If you want to KNOW that you can meet these requirements without distortion or compression, that gets a whole lot more difficult, and will require actual comprehensive verification, both of the system’s individual component behavior, and actual finished system capabilities at listening positions.

The methods of implementing the system factor in HEAVILY in the ability to meet output capabilities.

Loudspeaker sensitivity and linearity from baseline nominal to maximum.
Amplifier output capabilities, across the entire bandwidth, driving those specific loudspeakers, with real content.
Listening distance.
EQ as applied to playback curves.
EQ as applied to ‘correction’.
Personal subjective distortion tolerance and sensitivity.

Peak output capabilities alone do not guarantee sound quality.

While ‘Reference Level’ or RP22 recommendations may be interpreted as, or even occasionally as implemented as standards, there is no reason they have to be your standards. If you can determine what your own requirements are, you may be able to allocate resources to maximize sound quality in real use more effectively than you would meeting recommended requirements that may not apply to you.

But until you know, for sure, it will be okay. If it’s too quiet for you, turn it up. If it sounds too loud, for whatever reason, turn it down. If that approach works for you, run it!

If you can’t find a level that makes you silly happy, where you can hear everything, enjoy the full spectrum of the experience, comfortably, with ‘full immersion’ then there’s something wrong, and it’s worth figuring out. It could be maximum SPL, but it could also be something else. The Usual Suspect list is a long one.

The ‘try and see’ playback level method worked before ‘Reference Level’ was ever a thing, and it never stopped, which is a good thing, because content these days often makes it irrelevant.

For all the mathematics, standards, certifications, and marketing that surround this subject, the central question is surprisingly simple. What problem is Reference Level actually solving? It cannot be listener preference. The evidence suggests most listeners choose something else when given the opportunity. It cannot be industry practice. The evidence suggests theaters, streaming platforms, and content providers have all drifted from the idealized consistency the standard was intended to create. It cannot be a necessity. Millions of systems that will never approach Reference Level continue to deliver satisfying and emotionally engaging experiences every day. Reference Level remains useful. It remains measurable. It remains a valid engineering target. None of those things are in dispute. What is in dispute is the assumption that usefulness automatically confers universality. Somewhere along the way, a calibration procedure became a hierarchy. A tool became a destination. A recommendation became an expectation. The conversation shifted from "Does this system accomplish the listener's goals?" to "How closely does this system conform to the standard?" Those are not the same question. The first begins with the listener. The second begins with the standard. Knowing the difference matters because standards are not discovered. They are born from the need to define an obvious deficiency, and then lead that deficiency to a crafted resolution. However, they are products of assumptions, priorities, compromises, and constraints. They are models of reality, not reality itself. When a model repeatedly collides with observed behavior, the intellectually honest response is not to blame the observations, it is to re-examine the model. That is ultimately the challenge facing Reference Level today. Not whether it can be achieved. Not whether it can be measured. Not whether it can be certified. But whether the industry has mistaken a coordinate on the map for the destination itself. And if that question makes people uncomfortable, it should. Because the answer has implications far beyond a volume knob.

Amplifier Power and Speaker Output: How Much Do You Really Need?

A thesis on the practical matters of: Reference Level, RP22 developments, and further complications… Complete with Pictures.

Table Of Contents

Introduction: The Genesis Of Where We Are

System builders obsess over computer models and paper specs instead of asking what they actually need to achieve at the physical listening seat.