Are We Over-Optimizing Loudspeakers and Underestimating the Room?

[ahofer]

Room and loudspeakers are largely independent of each other. Good loudspeakers sound better than bad loudspeakers in any room. To quote Dr. Toole:


It is the same principal as music instruments and concert halls. You'll still recognize the quality of a top concert grand, relative to a mediocre piano, in a bad concert hall. You can't compensate for the differences between the top quality instrument and a poor one by moving to a better concert hall. Both the instruments (loudspeakers) and concert hall (listening room) are important.

This. A speaker that does well in the spinorama should make the most of a conventional space.

 

[Chris A]

Ummm...not so much as you might imply (I'm referring to two posts back.) See this "field report" on the effect of bad room acoustics that can't be overcome. This effect is even more pronounced (IME) in small rooms.

Even the world's best performing ensembles are at the mercy of room acoustics, it seems.

Chris

 

[olieb]

From the speaker design perspective we seem to put 99% of the emphasis on the anechoic performance, and expect it to translate perfectly to any room, while as mentioned in the recent thread on listening during speaker development, how the speaker actually performs in a room is a pretty complicated equation.

I fully agree.
Below the transition (150-200Hz) the situation is rather simple and therefore we can use an electronic bandaid and apply EQ.
This can “correct" the steady state in the listening zone of the room, depending number and positions of the low frequency transducers.

But above 500Hz the room does its thing. A neutral speaker with smooth directivity is always a good start, but if there is something wrong with the room (too strong reflections, uneven reflections, too early reflections, specular reflections, too long decay, uneven decay, ...) there is no "room correction" that can help, just "room treatment" (and to a certain degree, getting very close to the speakers).
Gladly, we can listen “through the room" and even more gladly, there is hardly a way of comparing rooms. So, we just get used to it. But there is a reason, why studios invest a lot into room treatment.
I recorded "my room" with in-ear microphones and ever since I know what I do not like about it. (Not that I did not have a hunch before.)
Toole's 4th ed has an interesting experiment about the relative impact of speakers and rooms on perceived sound quality (as experienced in competent binaural recordings) in Chapter 5.4/5.5.

 

[Torin Krell]

When putting together a top quality sound system these 5 steps in the following order work best for me (in agreement with much that has already been said in this thread):

1) Choose quality gear with accurate response.
2) Choose a listening room with acoustically reasonable acoustic dimensions and symmetry.
3) Optimize speaker and listening positions for best response.
4) Add effective well placed room treatment that maintains "envelopment" and works to treat even the low bass (this topic can take several chapters in a book all by itself).
5) Use PEQ or/and DSP to optimize response.

Steps 1-4 will at best make the corrections needed by step 5 quite small, which will also likely result in accurate sound in more listening positions throughout the listening room.

 

[Timcognito]

There are the most important elements that trounce all of those mentioned above after buying a well measuring speaker and placing them correctly in a room. So much that they are highly perceptible by all with normal hearing for ones age. Combining many of these technical elements result in outstanding listening satisfaction

• An engaging musical composition
• Accomplished musicians, musicianship and when added, singing acumen
• Proper recording techniques that include, microphones; number, type and placement
• The venue of the recording
• Engineering and Mastering highlighting or balancing elements SPL
• A non distracting listening environment

No playback device can do as much to realism and visceral involvement, and many such recordings do exist adding highly perceptible listening quality. The front end adds the most most important performance and technical aspects to audio.

 

[RexrothPigeon]

But I’m wondering whether we should shift more focus toward system-level optimization instead of treating the loudspeaker as an isolated object.

am i misunderstanding or is this what cardioids like D&D 8C are trying to do?

 

[Ype]

Or > From Spinorama to Living Room: How Much Survives?

I’d like to raise a question about system optimization that usually is seen here on ASR as a side issue at other Themes.

We put a lot of emphasis on designing loudspeakers to measure as close to “perfect” as possible: flat on-axis response, smooth directivity, low distortion, well-behaved off-axis curves. That makes sense, and I fully agree that a speaker should be fundamentally well-engineered.

However, in actual listening environments, the final result is heavily shaped by the room and by post-processing: room correction (Dirac, ARC, RoomFit), manual EQ, placement constraints, and acoustic treatment. In many setups, especially in normal living rooms, these factors introduce changes that are far larger than the residual imperfections of a well-designed speaker.

So one question is:
How much of the effort spent on achieving “near-perfect” anechoic performance actually survives in the final in-room result?

Or more provocatively:
Are we optimizing the right variables, or are we overfitting the loudspeaker itself while the dominant errors come from elsewhere?

So maybe the more relevant question is not “How perfect is the speaker in anechoic conditions?”, but:

“How robust is the speaker’s performance under real-world conditions and correction systems?”

Or, to put it another way:
After DSP and room correction, what’s actually left of that perfectly calibrated speaker?

I’m not arguing against good measurements – quite the opposite. But I’m wondering whether we should shift more focus toward system-level optimization instead of treating the loudspeaker as an isolated object.

Curious to hear your thoughts.

If you only use bass management as you should everything will be left of your properly designed speaker.
Your ears do not hear what a microphone measures above the transition frequency, so EQ'ing that with in room measurements is very, very bad idea.
Very bad. If you feel your speakers sound to bright or dull, it will be due to the recording and/or your person preference, and for that one uses tone controls.

I would highly recommend listening to a talk by Floyd Toole who will explain this much more elegantly then I ever could.

 

[Ype]

This is not to say that proper speaker placement and sensible listening positions are not important.
But you cannot compensate for a bad loudspeaker without ruining the sound in another way.
For example, dampening the reflections so much that you lose a sense of envelopment to try and compensate for terrible off axis response from your speakers.

This is why starting with a properly designed speaker is so important.

 

[Audionaut]

So: do everything that you can to find the minimum-phase response of the loudspeaker and equalize it to flat under quasi-anechoic conditions with the mic at 1m. Then place the mic at the listening position. Apply generous smoothing to the upper frequencies and then use that as the target curve. Alternatively, you can choose to forego MLP EQ of the upper frequencies altogether. The freqs below Schroder can be equalized as you wish.

Hi Keith,
You’re probably one of the people here on ASR who has delved most deeply into measurements, DSP, and room-tuning for speakers.
That’s why I’d like to ask you what you’ve found in your practice to be the most important aspects of room-tuning speakers using software like Accurate.
The frequency response will likely differ from the factory settings of the speaker. > Is a perfectly flat frequency response still the defining characteristic and prerequisite of a good speaker for you, or do you tend to think: This is a parameter that can be adjusted in practice?

After all, you can’t change everything using Accurate, for example: What can you change - and what remains, in any case, a typical, unchangeable characteristic of the speaker?

 

[Keith_W]

The frequency response will likely differ from the factory settings of the speaker. > Is a perfectly flat frequency response still the defining characteristic and prerequisite of a good speaker for you, or do you tend to think: This is a parameter that can be adjusted in practice?

After all, you can’t change everything using Accurate, for example: What can you change - and what remains, in any case, a typical, unchangeable characteristic of the speaker?

The way you phrase it, that is a huge question. It is easier to simply say this: you need to start with a good loudspeaker. You want to do as little DSP as you can get away with. I don't need to explain what a good loudspeaker is, go look at any of Amir's or Erin's speaker reviews and they show you where speakers deviate from ideal. You can not change anything physical about the loudspeaker, and most things about a loudspeaker's performance are already set in stone. About the ONLY thing you can change with DSP are the crossovers, driver linearisation, tonal balance, and low frequency room integration. DSP crossovers can significantly outperform traditional analog crossovers.

The most important thing you can not change about a loudspeaker is its directivity. Let's take a Martin Logan Prodigy loudspeaker, measurements from spinorama.org:

Looks pretty terrible, the on-axis is all over the place and far from smooth. OK so let's DSP it so that it's flat.

Fortunately, the reviewers did that for us They normalized the on-axis to flat. You can see that it's still pretty choppy off-axis, and both derivatives (ERDI and SPDI) remain the same, as they should be.

My view on this may be rather controversial on ASR: I don't think that smooth directivity is as important as it's made out to be. The aim is for the reflections and sound power to be spectrally correct, but most of what you hear is the on-axis response or maybe the listening window. The presence of ordinary room furnishings, asymmetrical rooms or layout will already distort the spectrally correct reflections. Thousands of people in our hobby don't have symmetrical listening rooms which are perfectly mirrored left to right, yet they can still enjoy the sound of their systems. I am NOT arguing that we should abandon smooth directivity as a design goal altogether, all I am saying is that slight imperfection, or even a lot of imperfection (as we see here) is not disastrous. OTOH it is the on-axis response that needs to be as perfect as possible, and that can be achieved with DSP.

 

[sigbergaudio]

@Keith_W I rarely disagree with what you write, but to use DSP to get a flat on-axis response on a speaker that isn't very good to begin with, rarely sounds natural to me. My apologies in advance if I misunderstood what you meant here.

 

[Keith_W]

@Keith_W I rarely disagree with what you write, but to use DSP to get a flat on-axis response on a speaker that isn't very good to begin with, rarely sounds natural to me. My apologies in advance if I misunderstood what you meant here.

Hi Thorbjorn, I think most DSP practitioners (including myself) have made the mistake of making the measurement look perfect without thinking about what we are correcting. In my experience, DSP lives or dies on the quality of the measurement. I have heard improvements in some very flawed loudspeakers, including mine. But I have also destroyed the sound without understanding why.

I should have caveated my statement by re-emphasizing that you need to start off with a good quality loudspeaker. If you try to make a loudspeaker do what it doesn't want to do, bad results will follow. To take one extreme: it is possible to make a small bookshelf speaker flat down to 20Hz. But you will lose about 50-60dB of headroom and have a tonne of bass distortion.

 

[Audionaut]

I don't need to explain what a good loudspeaker is, go look at any of Amir's or Erin's speaker reviews and they show you where speakers deviate from ideal

No, you don't have to explain; I just wanted to hear your opinion on it.
I've built a few speakers myself using the DCX2496, ADAU DSP, and Hypex board amplifiers.
Personally, I think the first step is the most important: choosing the drivers.
If they’re high-quality and you operate them within their specifications, you can experiment quite effectively with today’s technology, because the possibilities are endless and reversibel - unlike in the past, when physical resistors, capacitors, and inductors had to be swapped out on passive crossovers.

 

[Keith_W]

Personally, I think the first step is the most important: choosing the drivers.
If they’re high-quality and you operate them within their specifications, you can experiment quite effectively with today’s technology, because the possibilities are endless and reversibel - unlike in the past, when physical resistors, capacitors, and inductors had to be swapped out on passive crossovers.

You are right. I should also note that you can approach the limits of a speaker driver's performance more closely with a DSP crossover. When choosing XO points, the lower point is dictated by where the Fs (resonant frequency) is, and the upper point is onset of cone breakup / loss of pistonic behaviour. We traditionally choose the corner freq for the XO point to be 1-2 octaves above or below Fs/cone breakup depending on the order of the XO. But with DSP XO's, we can go much closer because they can be made much steeper with minimal cost. Try to price the cost of an 80Hz low-pass filter, you need expensive giant inductors, and many of them if you want a steep slope. Then you need a massive amplifier to heat up the XO network and send what remains to the driver With a DSP XO, it's a cinch.

 

[Audionaut]

In this thread

https://www.audiosciencereview.com/...-here-hate-all-very-expensive-speakers.71008/

we had some, in my opinion, very interesting—though unfortunately off-topic (Thread is closed) —discussions about bass reproduction and the influence of the room.

Among other things, the discussion centered on how to achieve a sound pressure level of 120 dB at 30 Hz with only 3% distortion—and whether this level of distortion at 30 Hz and very high volumes is even practical or realistically achievable in a "living room."

We also discussed the reversal of this principle—that is, when the room itself becomes the enclosure, as implemented here.

YT Video > A DIY builder creates a rotary subwoofer capable of producing extreme infrasonic bass frequencies that shake an entire house and demonstrate sound reproduction far below the limits of conventional speakers.

That's interesting, but I need to read up on the theoretical basics first to understand exactly what he's doing generally and especially with the propeller.

 

[thewas]

That's interesting, but I need to read up on the theoretical basics first to understand exactly what he's doing generally and especially with the propeller.

From what I understand such are changing the angle (from positive to negative values) of the propeller blades according to the input signal, thus pumping air in both directions like a membrane would.

 

[smygolf]

From the roatary sub wiki

Rotary woofer - Wikipedia

en.wikipedia.org

Instead of using a moving electromagnet (voice coil) placed within the field of a stationary permanent magnet to drive a cone, like a conventional subwoofer, on a rotary woofer, the voice coil's motion is used to change the angle of a fixed-rotation-speed set of fan blades in order to generate sound pressure waves. The pitch of the blades changes according to the signal the amplifier supplies, producing a modulated sound wave due to the air moved by the spinning blades