An Enticing Marketing Story, Theory Without Measurement?

[Blumlein 88]

This is absolutely the case, there is no standardization at all in studio design.

I can share some experience from the production side that folks here might find interesting.

I worked for a while in a studio designed by a top designer. We struggled for years because both the recording space and control room were unbelievably damped. It sounded nothing like any room I'd ever been in. It was designed to meet specific criteria for frequency response and reverberation times. But the space was small. The designer managed to meet the "paper specs" at the cost of a sub-optimal working space. It wasn't terrible, but there were some challenges. Eventually someone got frustrated and just took out some of the treatment down. It was a wonderful improvement.

There are aspects of recording/mixing that I find easier to do in a "normal" room, as opposed to a typical studio environment.

For example, I find getting the basic level balances between elements of the mix easier to do in a regular room. And this is the most important part of the mix! One would think balancing the relative levels, the voice, the drums, the instruments, would be easier on a nice studio monitor in a controlled space, but it's not that simple.

There is something that Floyd said about engineers using bad speakers while working that I would like to clarify.

Many engineers, whether recording, mixing, or mastering do use bad speakers in the process. But the use is not based on the theory that if it sounds good on these speakers it wIll sound good anywhere. That is simply not the case. Relying on a bad speaker when producing music will basically give you the inverse pattern of its deficiencies in your mix.

Checking the mix on the bad speaker gives the engineer information about the mix that is not always apparent on the main studio monitors. By checking on different speakers, the engineer can get a sense of how well the mix will translate to other systems in the "real world."

One of the shocking things when you start out trying to make recordings is that just making your mix sound good on your studio monitors does not suffice to make a mix that translates well across other playback systems. This is a complex subject. It's taken me years to get a handle on this, and it's still a challenge. (I've been working in pro audio 30 years of so. I'm a slow learner

I have a "cognitive framework" that I have come up with to help with all aspects of music production, and it addresses the issues with working on bad speakers, as well as the issues of working in imperfect studios. I've dubbed it:

The Principle Of Listening

In a nutshell, it's a suggestion that all changes that are made to a recording as it moves through the production process should be made based on how the change affects the sound of the recording. That means the change is made by ear. This sounds ridiculous at first, because it seems obvious. What else could you do?

It turns out that anyone making a recording can be easily led astray and make decisions about the signal on criteria that are not based on how the change effects the sound of the mix.

For example, music is produced mostly on the computer these days, and there is a lot of onscreen visual stimulation. A listeners attention is inevitable drawn to focus on the screen, which has a profound effect on auditory perception (and not a good one.) Even if you are working and trying to ignore the visual "noise" in the workspace.

Another common side-track is to make decisions about which piece of equipment or software to use based on its reputation: Oh, so and so famous mixer uses this DSP compressor plugin on every mix! So I'll use that on all my mixes too!...Or... this mic is advertised as being great for vocals, so that's the ticket to making our smash hit record!

Another common example is the use of spectrum analyzer to assess your mix. A tool like this can tell you where you might have problems in the mix, but at best these serve as clues. You cannot fix the issue shown on the frequency analyzer by just making changes with global EQ until the shape of the frequency response looks like what you are going for. Such an approach will mess up the artfully created musicality of the mix.

So the frequency analyzer can give you information about your mix. But how does the engineer/mixer use this information?

This same issue presents itself when you check your mixes on other speakers. For example, a practice with a very long pedigree is checking the mix on a car audio system. Say you listen in the car, and the mix sounds totally off! You can't hear certain instruments, it has boomy low midrange, uneven dynamics! Very disappointing to everyone working on it.

How do you fix these newly perceived issues? Suppose you could bring your mix system right out to the car, plug it in to the audio system and fix the problems right there. (And you can do this now-a-days because it's trivial to mix on a laptop.)

But if you sit in the car, making adjustments to your mix until it sounds great, you will have made no progress, and probably made your mix much worse! Such an approach to mixing will result in a mix that will probably not translate to other systems at all.

To use the Principle of Listening you have to be able to perceive the issue you noticed on another system on your main studio monitors. And if the monitors and room are at least decent, you usually can. You can hear those very same issue on the studio monitors, but it will take concentration and critical listening skills.

This paradoxical issue is exacerbated in part because modern studio monitors have great dynamic range, and are utilized in the overly deadened spaces Floyd mentioned.

In a dead room, this "perceptual dynamic" range is enhanced far beyond a regular listening environment. Such a monitoring environment can be so clear, that it becomes very hard to judge relative volume of signals. You can mix a loud instrument with a soft one, and both will be perfectly perceptible in the studio. But in an average playback environment, the loud signal can "mask" the quite one dramatically, so the balance of the instruments is way off.

A regular room is more reflective and the sound is more diffuse. And the playback system is usually not so great. In such a space, the sound interacts with the speaker, reflects off objects, and energizes the physical space. This general "mixing up" of the sound makes the difference in the average sound energy level of different sounds in the mix much more apparent.

So problem for the engineer is that they need to fix a problem in their mix that is not a problem on their main monitors!

The first part of using the Principle of Listening is to tune into perceiving the problem you heard in the car, for example, on your main monitors. You do this with critical listening. Once you can clearly perceive the issue in your monitors, you can make adjustments that correct the problem by listening, making sure the mix still sounds good while making the adjustment. Not just "blindly" changing something in hopes it fixes it.

Bass response is one of the biggest problems for monitoring environments, and the modern, dead-ish control rooms can absorb a lot of bass frequencies. This can result in mixes that are way too heavy in the low frequencies, which can be very disheartening. This is done to give the room a more even frequency response, but it comes with all sorts of costs.

Another big issue is that overly treated rooms, coupled with good monitors, can handle much bigger dynamic ranges than everyday systems. Our ear is not that sensitive to changes in levels. Most of the cues in a mix that communicate "loudness" are delivered by timbre, not level.

Another way we perceive dynamic sound level changes is by how the sound energy is transferred to the objects in the room, including the speaker itself, the walls, the floor, things that rattle. These level changes can be almost invisible in some control rooms. The speakers are very isolated from the structure of the room, the reverberation times are reduced (with attempts to maintain neutral frequency response), the amps for the speakers are very powerful and won't distort.

So some person out there is listening to your mix, sounds pretty good! But it has some very large low frequency dynamic hits, that on your studio monitors played just fine. The listener decides crank up the volume, and all of a sudden the speaker is distorting, the media cabinet is resonating, things are falling off the walls. Not so fun for the listener!

If you know you have an issue with some frequencies or dynamics, with critical listening, and some helper tools, you can address the problem. Making the changes by ear allows the engineer to gauge the type and amount of adjustments to while making sure those changes at the least don't detract from the musicality of the mix. Ideally the fixing "problems" in a mix ultimately results in better mix all around, no matter what system its played on.

Despite these observations, the thought of treating production environments more like regular listening environments gives me pause. As I've described, some issues in a mix can be much easier to perceive in a "real world" setting.

But...

Studio monitors enable a level of clarity that is essential in making the nuanced adjustments that a good mix needs.

The biggest problem is perhaps the challenge of accurately assessing the amount and color of the ambience in the recording, because the ambience of the working space will blend with the ambience in the recording, preventing you from hearing what you really have in your mix. It can lead to similar problems of mix translation.

This could be less significant in the mastering process, as the changes to sound there are usually broad changes to the whole mix, and one of the main goals of mastering is helping the mix translate well once it's released to the world. So working in an environment that more closely resembles real world listening environments might have a lot to recommend it.

The goal would be to try and standardize productions environments, and finding a happy medium between "live" or "dead" sounding spaces. This could probably be done, while keeping some of the qualities of the room that make it relatively "neutral" in it's overall sound.

I think it's a great idea, though it probably won't happen. Maybe at the highest levels of the audio post production system something like standardized room treatments can emerge. I think this practice is used to a degree for final mixes for movies. But a lot of music production is done in personal spaces, or small commercial studios that have been set up by an individual or small team. These endeavors are severely limited in budget. Yet they do contribute tremendously to the musical content that makes it out to the world.

The reason this problem is not more urgent is that the musical products we hear on a day to day basis, at home, in the car, playing in a store, wherever, are produced by highly skilled craftsmen, who have learned to create mixes in imperfect environments that translate well.

I have some understanding and experience of what you are describing here. I still think the situation, which is as you describe it, is a gigantic amplification of the Circle of Confusion. You mention movie mixes, and while not always great, and only a modicum of standardization takes place, that has decreased the Circle of Confusion. Movies until recently also had to mix for a smaller range of likely listening environments.

When I listen to most current recordings highly skilled craftsmen is not the description that comes to mind. On the other hand, having worked with some recordings, yes those people do have real skills at mixing things. At least the few I've been around do need to get clued in that you don't have to squash it to death. When you can get them to back off on that a little their skill is even more apparent. Good things I've no skill to do myself. I suppose if not for their skills the squashed to death recordings would be far worse. But that smacks of a Sisyphean task to toil at to me.

I get that things aren't going back. I get at least a little why they are that way. I remember making a nice cassette dub of some of my favorite symphonies. My car was on the quieter side of average, but you couldn't enjoy those in the car. At least half the music was simply inaudible. Which is a variation on unlistenable, because you can't listen to what you cannot hear. Careful compression can make that kind of recording close to listenable in the car. But it can't ever really sound right. So yes even pop music needs some help for such environments. But I'm trying to figure out what environment most current recordings are for. They are squashed beyond what is needed in a car. Do that many people listen while working in a place as noisy as a steel mill? Some friends had an all acoustic Christmas lullaby returned to them with a DR6. You know slightly more loud than Metallica's Death Magnetic. I don't know anywhere such a mastering and mix translates. A couple of the musicians said they could not hear anywhere that sounded like their instrument.

I'm sorry for venting. I'm of course not blaming your personally. I think your description of the situation is accurate, and honest.

 

[RayDunzl]

I figure one reason for the squashing of recordings against the limit is because "they" are looking at a linear instead of a logarithmic display when observing the waves.

-30dB looks like nothing on a linear display, still has some meat on a logarithmic display

1000Hz, 0dB, -10dB, -20dB, -30dB ---- 60 or more dB remaining to be filled inside that straight line at the end...

That -30dB wave on the linear display is just not very impressive, is it?

---

Ok, 1kHz was a bad choice, unexpected interaction with the display, so here's 50Hz (display makes more sense), 0dB, -10dB, -20dB, -30dB, and -40dB:

Anyway, that's my theory. The wave doesn't look manly enough unless it fills the linear space available.

---

Here's a rather dynamic recording:

That's an hour of music above...

 

[Cosmik]

Doing a 'meta analysis' of the comments it does often seem that people go down a rabbit hole of room treatments and 'room correction' based on laptop-and-mic measurements that they assume they must be hearing, but later it is shown that they were simply not hearing what the laptop screen was telling them.

And in all of these cases, any anecdotal "vast improvement" could easily just be a novelty that lasts for a limited time until the next vast improvement comes along.

 

[Bjorn]

Well it seems that you have not read my book because, there is evidence that humans do not "hear" the ringing (European research not mine - see Section 8.3). It is counterintuitive, but so are other aspects of psychoacoustics. Yes, multi-sub solutions only work in the sub frequency range, but the most energetic room modes tend to be below 80 Hz, and these can be very well controlled. Above 80 Hz we are into the adjacent boundary region (Chapter 9) and equalization is effective because this is a sound power radiation issue and it should be used certainly up to the transition frequency, and sometimes a bit higher - 400 -500 Hz. BTW the Schroeder frequency calculation is for large reverberant auditoriums, not small listening rooms - his definition, not mine - see Section 6.1.

You find studies on many matters that aren't necessarily correct. An example is a study by european broadcasting companies that showed that MP3 in 128 Kbps was inaudible compared to lossless. But the fact is that a trained listener can detect even MP3 in 320 kbps from lossless.

I can't comment on that research, but IMO the perceived difference between cleaning up the long decay and resonances in the 80 Hz to 250 Hz area is enormous. To say this isn't audible is completely baffling to me. There must be something seriously wrong with that study. We are not talking minor differences here. The tightness, the experienced of "suddency", feeling of slam and impact, and how much cleaner the music is has been dramatically improved. Anyone who has experiences this know this well. Obviously you can say this is subjective nonsense with no proof, but does it really make sense that this area should be totally inaudible to long decay and strong resonances?

Using EQ doesn't work well unless it's minimum phase behavior. While there might be a few regions above 80 Hz that are close to being minimum phase, most aren't, hence I don't agree with you here that EQ works well. The difference between a setup that is based on EQ in this region and quality room treatment is night and day IMO.

Setting up a multi-sub system is straightforward in a rectangular room (Welti papers and Section 8.2.6) but Sound Field Management, if you have access to it (JBL Synthesis), is also straightforward, but requiring some measurements and the optimization algorithm. The results are predictable Section 8.2.8), and improvement is possible in any room we have yet encountered. Without SFM or a functional equivalent, you are right that setup takes a lot of trial and error.

I have set up multiple subwoofers several times. Don't get me wrong, I think it works well to some degree but it also has its compromises. For once and as mentioned it only works in sub frequencies, leaving the bass and low midrange frequencies still very muddy. It seems you believe this is inaudible based on the study you mention and here we simply have to disagree.


Secondly, it can easily sound a bit messy compared to having the subwoofers in front of you. This depends on several factors but if the room is rather small and you place subwoofers in the rear with a crossover at 80 Hz they will often be localized.

And the seamless integration you get from having subwoofers in the front close to the front speakers just isn't the same with multiple subwoofers spread around the room with large distance between them. Even with signal alignment. That is of course my experience, and you may disagree.

You said: "Accuracy requires attenuation of early arriving reflections no matter how the speaker measures off-axis." Now we part company. Stereo is mono left, mono right and double mono amplitude panned for all images across the soundstage. To this basic set of images is incorporated some amount of poorly correlated real or synthesized reflected sound in the two channels to create a sense of space. In studio mixes different instruments and voices of the mix can be put in different spaces. Stereo is not an "encode/decode" system, it is merely a two channel delivery system and there are enormous variations depending on how the channel information is utilized in the miking and mixing processes. In many stereo recordings sound will be heard emerging only from the left or right loudspeaker, in which case the sound quality evaluations are very similar to those in mono presentations. Section 7.4.2. In those cases there is evidence that listeners prefer to have a sense of space around the point-source loudspeaker. It seems that you prefer "pinpoint localizations" - and you have company, but it is not the only preference. Chapter 7 includes a lot of information on the effects of lateral reflections on image location and other aspects.

It's more than just "pinpoint localizations" while that's certainly a part of it, and good localization is very much related to accuracy of the mix/mastering. What speaker that measures great off-axis does is giving us some less variation in tonality. But only to a certain extent. The complex impedance of boundaries and furnitures still give us something different. And the fact that signal arrives later than the direct signal has its effect alone. Both clarity, intelligibility and partially tonality are also highly improved with attenuation high gain reflections/comb filtering effects.

IMO you draw the conclusion wrongly when you this has to be imply lack of spaciousness. That's absolutely possible with treatment as well. Actually, you can have more of it compared to no treatment.

Something interesting here is how close a certain design principle applied to a small room is in the time domain (ETC measurement) is to the best concert halls. What you have is a time span (called ITD or ISD) of strongly attenuated early reflections and preferably to a point where it's inaudibly, followed by diffused sound that exponentially decays. If you compare a small room treated this way with a well designed concert hall they are very similar. That means they can also be close in terms of a spacious sound field or how large the room is experienced.

I think I describe in the book, certainly elsewhere, that very early in my experiments I set up a room with heavy lined drapes that could be moved forward or back to vary the amplitude of the side-wall reflections. Among my group of volunteer listeners there was one who was sufficiently enthused to set It up in his personal room. He concluded what was quite clear from our in-house evaluations (which were not short term sessions) that preference was significantly program dependent. There was a "classical" setup and a "rock/pop" setup.

As mentioned earlier; it's crucial that the treatment is broadband for a good result. Drapes (even heavy) will not treat specular reflections evenly or broadband. They will function as a filter, only absorbing some of the very highs of the reflected energy. And they may not absorb the highs very well either. This is exactly the type of treatment that leads to a worse result in many ways. Let us not use poor examples of treatment to conclude what acoustic treatment does. Quality treatment gives a complete different result just like a quality speaker does.

In the end I will offer that unless one begins with fundamentally well-designed loudspeakers no amount of Room EQ or acoustical fiddling will save the day. With good loudspeakers the job is much easier, and intelligent manipulation of acoustics devices may or may not lead to greater satisfaction. Bass MUST be fixed and that includes adjacent boundary effects below the transition frequency.

But so long as one stays stuck in the stereo groove, there can not be one physical solution for all listeners. It is inherently a directionally and spatially deprived format, with everyone seeking a "solution", when there is none that does not involve multiple additional channels. I live with 9.4.6 and use them all from time to time.

I certainly agree that one should always start with good speaker designs. I also believe using separate subwoofer(s) is beneficial in most rooms.
On other areas we disagree. Multiple channels, while giving increased spaciousness, also creates a more chaotic result due to lobing (space between sources) and comb filtering. I personally don't like the result of that. I'll rather have spaciousness without these issues. Something that is possible with the right choice of treatment.

 

[Bjorn]

Oh, I see people do this all the time on forums. The most ardent advocates of "block side reflections" come back a year or two later and say: "I removed all of that and was surprised how much better the sound was!"

I'm sure you have both, but my general experience after working with acoustics for years is the opposite is the more the norm. It's also heavily depended on type of treatment one uses, exactly how it's being used and the overall treatment/result. Many fall into the trap of very band limited treatment and placed wrongly or uses too much of it. This certainly leaves to a poorer result in several areas.
As mentioned, there are several ways to treat reflections. Absorption isn't the only way.

 

[Bjorn]

One of the best experience I had was setting up my Revel Salon 2 speakers in a bare highly reflective room and playing classical/orchestral music. It sounded amazing. I did have to modify that once I listened to rock and pop music and did so with usual furnishings. This is a living space and acoustic products have no business being there.

At madrona we have a highly treated theater room. It is for multichannel playback. Here is a shot of it:

View attachment 25320

There is about $25,000 worth of acoustic products in there, some that we built and some that we bought off-the-shelf. The room was designed by Keith Yates. What you see are all behind acoustically transparent fabric that then is backlit using LED lights. There is a lot more on the other surfaces and the ceiling. It all sounds incredible.

However, when we put even high-end stereo speakers in there (e.g KEF Blade 2, Revel Salon 2, etc.) they all sound extremely dead and dull. I remember the KEF people who brought the Blade 2 and after hearing it in this room, they wanted to slash their wrists. I agreed and we took them out and put them in our larger, untreated space and they sounded far better.

So no, I am 100% with Dr. Toole here. Fancy acoustic products is not a ticket to excellent sound. And reflections are not the thing we should fear. Too much is a problem and bare rooms like above do need some but there is no measure of more the better.

There are of course exceptions. My chief designer loved both the KEFs and Salon 2s in our theater. He mixes and records music so likely has adapter his hearing to a) hear the effect of reflections more ab) prefer them that way. Not so for most of us.

Dead and dull normally means overdampening and especially of the higher frequencies. We shouldn't fall into the trap that because we have heard a certain treated room, even it was expensive, that gives us the full answer to treatment. There are many ways to treat a room.

I don't know other than what I see in the picture, but what I see is treatment that isn't broadband (very shallow scattering devices), not placed optimally in terms of where you actually get high gain reflections, products that doesn't diffuse well and the combination of different scattering units placed adjacently indicates lobing issues. Much of that treatment (seen in the picture) is barely even seen by anything other than the highest frequencies! And when they are seen by the incident sound, are reflected to the floor and ceiling (2D redirection).

I can certainly understand why this room didn't perform that well with some speakers. But it's not because treatment generally doesn't work well. You should hear a better treated room

 

[jazzendapus]

I want to bring this back to the original topic.
When DRC such as Audyssey does its magic, what kind of measurements does it really rely on? It looks like it calculates distances from speakers to listening position(s) very accurately, and that's probably a sign that it can also do decent gated measurements, eliminating the energy from reflections and getting a pretty clear picture about how the speaker itself behaves. And if this is the case, doesn't it mean that it can correct actual speaker resonances (if there are) reasonably well?

 

[Bjorn]

Some of the debate is really about preference vs accuracy.

I find preference difficult. While psycho acoustic researchers here are interesting I also find they don't give definite answers. The reason is that there are so many aspects that come into play in regards to preference. One being the music material. Another being how the room is treated as a whole. It's not surprise that most will prefer some side wall reflections in an anechoic chamber where the brain is longing for some contribution or in a room that's treated with much absorption on the rear wall. No study has for example taking into account the use of quality diffusion in either the rear of the room or on side walls when looking at the preferences. On other words, they are lacking.

Another issue for me in regards to preference is the listening time span. I have many times found out that what impresses me right away is not what I prefer over time. So bottom line is that I don't consider researchers on preference to give accurate scientific answers. While they may give a clue, people have to test that for themselves in their environment over time.

With accuracy it's very different. Here we can base on something that measurable combined with our perception. If we look at high gain reflections it's no doubt it will color the sound. But exactly how will vary. The time, gain and "EQ" of the reflected sound depends upon the boundary impedance (how it effects the frequency composition of the reflected signal) and hence the resultant superposition of the reflected with the direct signal.

There isn't one effect. The time differential and the gain relationship, plus how the indirect signal is or is not modified by the boundary (its acoustical impedance that works on the frequency composition just like resistors, capacitors and inductors in a filter) results in varying perceptions. It can range from reinforced intelligibility to late arriving signals that result in an echo.

Aside from how the timing of the superposed reflections modifies perception of the superposed signals, the modification of the indirect signal by the boundary further modifies the timbral makeup and further changes it from the direct signal.
So there isn't simply one distinct change, it's like a continuum of perceptual variations depending on the combinations and permutations of the resultant superposed multiple signals. However, we do know very well that it has an audible effect and if accuracy is something one desires, treatment is absolutely necessary.

Multi channel is for example not an accurate presentation. The problem is that the various channels use gain and not accurate time relationships, they are simply delayed. So it's a hodgepodge of room reflections from multiple sources and then additional delayed direct sources that further confuse the arrival time relationships of all of the various arriving sources. Just make it so seemingly random that it seems bigger, but without any accurate imaging or localization or even intelligibility, but its "immersive" and "big" sounding.

However, I'm not going the debate with preferences. You can't really debate with someone who says he want this and that. What I'm mainly debating is accuracy. Pick what you want but also try out several things.

 

[Cosmik]

With accuracy it's very different. Here we can base on something that measurable combined with our perception. If we look at high gain reflections it's no doubt it will color the sound.

Why do you say "No doubt..."? And what is 'the sound'?

Where does that lack of doubt come from? I suggest it is because of what the laptop screen shows and takes no account of what the ears/brain are doing.

And unambiguous accuracy can be achieved by listening in an anechoic chamber. But using the laptop screen to adjust 'the sound' of a system in a real room in order to better resemble certain aspects of the anechoic measurements is illogical.

I think we need a new term: 'laptop sound'. It's the sound you get when you set up a system visually based on what is displayed on a laptop screen.

 

[Bjorn]

Why do you say "No doubt..."? And what is 'the sound'?

Where does that lack of doubt come from? I suggest it is because of what the laptop screen shows and takes no account of what the ears/brain are doing.

And unambiguous accuracy can be achieved by listening in an anechoic chamber. But using the laptop screen to adjust 'the sound' of a system in a real room in order to better resemble certain aspects of the anechoic measurements is illogical.

I think we need a new term: 'laptop sound'. It's the sound you get when you set up a system visually based on what is displayed on a laptop screen.

Did you read the rest of my answer below? It explains why. The audible effect of comb filtering is also well documented, but as I pointed out:
There isn't one effect. The time differential and the gain relationship, plus how the indirect signal is or is not modified by the boundary (its acoustical impedance that works on the frequency composition just like resistors, capacitors and inductors in a filter) results in varying perceptions.

Listening in an anechoic chamber is as correct as one might first think. It's an environment experienced as dead and not natural to the brain. That is exactly why the studios moved away from this approach.

 

[Cosmik]

Listening in an anechoic chamber is as correct as one might first think. It's an environment experienced as dead and not natural to the brain. That is exactly why the studios moved away from this approach.

Yes, I hate being in a dead room (after the initial novelty has worn off perhaps), but what I am saying is that the dead room gives unambiguous "accuracy" in that what appears on the laptop screen is the same as the unadulterated recorded signal (assuming a speaker with basic on-axis accuracy).

The inconsistency comes when people insist on listening in real rooms (because it sounds better), but aim for measurements that visually resemble the dead room's measurements. An example is the speaker with modified super-narrow dispersion that gets you that much closer to the dead room's measurements and sound if you are sitting in the right spot. It's where measurements take over from simply listening, and enjoying the sound of a room.

 

[RayDunzl]

I think we need a new term: 'laptop sound'. It's the sound you get when you set up a system visually based on what is displayed on a laptop screen.

I have desktop sound.

 

[Floyd Toole]

There is something that Floyd said about engineers using bad speakers while working that I would like to clarify.

Many engineers, whether recording, mixing, or mastering do use bad speakers in the process. But the use is not based on the theory that if it sounds good on these speakers it wIll sound good anywhere. That is simply not the case. Relying on a bad speaker when producing music will basically give you the inverse pattern of its deficiencies in your mix

This is a good discussion topic. I truly understand the need for different "views" of a work in progress. Traditionally this has been done using a collection of small loudspeakers littering the meter bridge. In this circumstance the listener is in the flawed acoustical near-field of the loudspeaker and the reflecting work surface, which simply adds another dimension to the flawed sound. The main point, however, is that the sound field is dominated by direct sound (including the console reflection). This being the case, to simulate different flawed loudspeakers one can simply modify the spectrum, the frequency response of the signal feeding a neutral loudspeaker. However, because the dominant flaws in loudspeakers are resonances this imitative EQ must be done with high frequency resolution, say 1/20-octave not 1/3-octave, in order to simulate the timbral changes. Obviously one could invent one's own spectral modifications to address audible features of interest. All of this could be digitally implemented and stored, accessible by icons - including a bypass icon.

One could have an infinity of virtual "loudspeakers" to test the mix on, without cluttering up the control room with lumps that do no favors for the ideally timbrally neutral main monitors. The most obvious differences between good speakers and mass market speakers is that small inexpensive units cannot play as loud, and have less bass. Nowadays some small bluetooth speakers sound quite passable - better than expected. The difference is that they are all active speakers with built-in EQ, designed by engineers who do measurements. A few active monitor (e.g. Barefoot) loudspeakers attempt "imitations", e.g. the "old school" NS-10, but I don't know how well it has been done. To complete the illusion one may need to find a way to change the color of the woofer cone to white .

But a significant barrier to such a proposition is that so many people still don't believe that measurements can describe what we hear. They think that cone material, box shape, brand name, user accolades, and on and on, are determinative. No doubt they matter, and some may contribute to the sound quality, but the contributions are measurable in an anechoic chamber and can be presented for visual inspection, for example in the spinorama format. This is now happening in a very limited way in the consumer loudspeaker world. However, very few professional loudspeaker manufacturers reveal anything of value to describe the performance of their products - even when asked. This needs to change.

 

[Kal Rubinson]

They think that cone material, box shape, brand name, user accolades, and on and on, are determinative.

That influences the information that is offered to the public in the form of advertised specifications. Much of what is offered can be sussed out by the eye or with a multimeter but acoustical information is sadly lacking........................in general. I have been in touch with a speaker company which even advertises that they have an impressive facility for acoustical testing but admit that they do not release that information to the public.

 

[Kvalsvoll]

150Hz to 200Hz is beyond the normal usage range of subs. I would not use them within that range. Normally its below 80Hz due to the inability to localise below that frequency. I would suspect the majority of significant room modes for most domestic listening rooms are below 100Hz. I use Acourate FIR DSP for the crossovers so the time domain adjustment is also catered for

Best crossover frequency depends on speakers and subwoofers/bass-system, and room. I usually recommend starting at 120hz, much higher starts to affect placement of bass instruments, and then you need to run the bass-system in stereo, which may not be supported by the processor in use.

The range 80-200hz is typically problematic, because there is not sufficient absorption at those low frequencies, and speaker-room is no longer minimum-phase. Look at frequency response, decay and group delay in this particular range. DSP can improve things, but it will never be as good as fixing acoustics first, end then apply dsp processing afterwards.

Keep in mind I am not talking about the subwoofers you find in the typical shop now, think custom installations and bass-systems dimensioned and selected for that particular install.

 

[March Audio]

Best crossover frequency depends on speakers and subwoofers/bass-system, and room. I usually recommend starting at 120hz, much higher starts to affect placement of bass instruments, and then you need to run the bass-system in stereo, which may not be supported by the processor in use.

The range 80-200hz is typically problematic, because there is not sufficient absorption at those low frequencies, and speaker-room is no longer minimum-phase. Look at frequency response, decay and group delay in this particular range. DSP can improve things, but it will never be as good as fixing acoustics first, end then apply dsp processing afterwards.

Keep in mind I am not talking about the subwoofers you find in the typical shop now, think custom installations and bass-systems dimensioned and selected for that particular install.

The typical 80hz crossover for home theatre was chosen for a reason, it's where the the likelyhood of localisation becomes small. I don't see a reason to disregard this.

I agree with you on the point that effective absorption is required within that region and my room is indeed treated to specifically deal with that range, however we were talking about subs. Dsp is still useful in that region. Where and why do you define the point of non min phase behaviour?

 

[b1daly]

This is a good discussion topic. I truly understand the need for different "views" of a work in progress. Traditionally this has been done using a collection of small loudspeakers littering the meter bridge. In this circumstance the listener is in the flawed acoustical near-field of the loudspeaker and the reflecting work surface, which simply adds another dimension to the flawed sound. The main point, however, is that the sound field is dominated by direct sound (including the console reflection). This being the case, to simulate different flawed loudspeakers one can simply modify the spectrum, the frequency response of the signal feeding a neutral loudspeaker. However, because the dominant flaws in loudspeakers are resonances this imitative EQ must be done with high frequency resolution, say 1/20-octave not 1/3-octave, in order to simulate the timbral changes. Obviously one could invent one's own spectral modifications to address audible features of interest. All of this could be digitally implemented and stored, accessible by icons - including a bypass icon.

One could have an infinity of virtual "loudspeakers" to test the mix on, without cluttering up the control room with lumps that do no favors for the ideally timbrally neutral main monitors. The most obvious differences between good speakers and mass market speakers is that small inexpensive units cannot play as loud, and have less bass. Nowadays some small bluetooth speakers sound quite passable - better than expected. The difference is that they are all active speakers with built-in EQ, designed by engineers who do measurements. A few active monitor (e.g. Barefoot) loudspeakers attempt "imitations", e.g. the "old school" NS-10, but I don't know how well it has been done. To complete the illusion one may need to find a way to change the color of the woofer cone to white .

But a significant barrier to such a proposition is that so many people still don't believe that measurements can describe what we hear. They think that cone material, box shape, brand name, user accolades, and on and on, are determinative. No doubt they matter, and some may contribute to the sound quality, but the contributions are measurable in an anechoic chamber and can be presented for visual inspection, for example in the spinorama format. This is now happening in a very limited way in the consumer loudspeaker world. However, very few professional loudspeaker manufacturers reveal anything of value to describe the performance of their products - even when asked. This needs to change.

There is another factor at play in how music producers/engineers work. Many (most) rely on being able to switch monitors at the workstation. (The console has gone the way of the DoDo, everyone works in front of a computer now, which is a “mixed bag”)

But mixers do hear their work in different environments, whether as part of a work process or happenstance. The observations feedback in a continuous process of learning over years.

To get a good mix it is essential to have a sense of how it will sound across the infinite variety of playback environments. This is an impossible task, but one makes progress.

I have only been introduced to your fascinating and vast body of work in the past few months.

I haven’t read your book, but watched some of your lectures, read a paper, and followed some discussions here enough that I think I have the gist of re: carefully done listening preference testing being highly correlated with the spin-o-rama measurements.

These results frankly blew my mind!

This is a vast and complex subject, but I got a little annoyed (not in a bad way, just set me to thinking) with some folks here who proclaimed that your results proved that evaluating speakers can be done under a similar framework as the “values” of this website.

Which are, loosely, that audio gear can be tested by measuring. The closer the output signal is to the input, the “better” the piece of gear is (in audio performance).

So I perceived this attitude that, duh, the best speakers are the most “accurate” with that being defined as the ability of the on axis speaker response to generate an acoustic signal that when captured by another transducer most closely resembles the input signal. And incorporating the relative accuracy of off axis sounds as the attenuate.

While I accept this result, I still find it surprising, and it calls out for an explanation.

One reason is that, if my understanding of the listening tests is correct, the listeners should have no reference point as to the nature of the original signal. So I see no presumptive reason why “accuracy” of the on-axis response would be preferred.

I had a couple of musings as to what could explain this:

It could be that despite the unbelievable amount of signal processing done to a modern recording, there are still enough acoustic cues present that allow us to perceive “deviation” from how things “should” sound.

The influence of off-axis sound is easier to understand.

Another idea was that speakers as sound producing devices just “sound better” when it happens that they have accurate on axis response. This doesn’t really answer the question, but moves it away from a narrow conception of audio reproduction “accuracy”.

Another explanation is that there are subtle ways for “out-of-band” information to influence the result.

The list I saw of the material you used to test the speakers had Tracy Chapman, Jennifer Warnes, and some other samples I can’t remember. But what struck me is that all of these are excellent recordings. Basically these were done by the people in the absolute top tier of skill level.

As I was going on about before, the best producers/engineers/mixers put tremendous effort at both the conscious and unconscious levels to make mixes that “translate” really well. This is a mysterious quality. There are objective aspects to it, but in some ways I feel there is a “musical” signal present in a good recording that operates in a different realm than what we think of as audio performance.

As long as the playback system is “good enough”, the “meaning” of music will be almost completely communicated.

There are a lot of factors both biological and cultural that could be at play here. But in some profound way, certain recordings “sound right” almost anywhere they are played.

This means the listener has an internal reference for how something “should sound”. The music creators work at this domain level, and use whatever monitor system the have hand to perceive the creation. And likewise on the listening end.

For most people, the best producers are so skilled at creating “high meta-signal level” that allow people to have powerful experiences on an incrediblely wide range of systems.

Perhaps there are sort of cultural musical archetypes that provide a built in reference point for how something should sound, and the best productions hew closely to those.

How music communicates meaning is very complex. But it could be that these meta-archetypes, operating across diverse perceptual and cognitive functions, can utilize “accuracy” to a degree. A producer cannot rely upon this, since most playback systems suck, but in their multi-faceted attempts to reference these “audio-archetypes” they do access components that are communicated better by “hi-fi” systems.

To anyone reading this who thinks this is just mumbo-jumbo, the best example I can think of is to imagine your favorite song. Your memory is probably not how it sounded on a particular system. Instead, especially if you are quite familiar with it over different systems, you have sort of sonic-image-memory that is independent of any specific playback.

Whatever combination of factors the artists combined to create this artistic product, an accurate system reveals the aspects of this that rely on high fidelity reproduction. So even though you don’t need to hear your favorite song on an amazing system to appreciate it, a great system will reveal subtle parts of the “music signal” enhancing your enjoyment.

Part of reason it works has to be because on the other end, the music creators are relying on monitors that are pretty good, and some details they pack into the mix are things that relatively good systems are required to hear.

Familiarity with the test material might also influence the results if the same recordings were used repeatedly in a test. A listener could “average” across the sets of speakers to get a sense of how the mix works”should” sound.

If so, this should show up in results.

It’s “on my list” to read your reference book and learn more about the discovery’s of yourself and colleagues.

Thank you for sharing your insight with us!

 

[edechamps]

I have been in touch with a speaker company which even advertises that they have an impressive facility for acoustical testing but admit that they do not release that information to the public.

That reminds me of that comment I made 5 years ago where I remarked that even speaker measurement "evangelists" Sean Olive and @Floyd Toole, who work (or have worked) at Harman in high-ranking acoustic engineering positions, were apparently unable to convince their own company to publish measurements for the speakers that they produce. That points to a strong disconnect between the marketing department and the science behind the sound of the speakers they sell.

 

[Kal Rubinson]

That points to a strong disconnect between the marketing department and the science behind the sound of the speakers they sell.

Indeed. The disconnect, of course, extends to through to most dealers, most users/purchasers and, sadly, to the perspective of most reviewers.

 

[Purité Audio]

I got ‘kicked off’ for posting some in-room measurements which the manufacturer didn’t enjoy, it makes one rather cautious!
Keith