Pro Audio Folklore

[Palmer]

On a more serious note- as I am thinking about the possible set up of a test bed, to see if similar, but different approach equipment does sound different - measurably, that is...
Since I have two each of the mobile mic preamps, the PSC Alphamix (I think is better sounding) and the Shure FP 33 - both battery powered, maybe there is a way to come up with a test chain, and I could ship one of each to Amir. Both do take line inputs, though that somewhat negates the outcome if a studio grade condenser is fed into it.
Since both don't do anything else than gain to the input, maybe that would be a good way to get into it?

 

[Cbdb2]

'll repeat what I said in answer to an earlier comment by @kemmler3D: "These aren't easy devices to test. If the sound is different, the transfer function will be different." I don't know if that's clear or not. The transfer function defines the totality of what happens to a signal between input and output.

transfer function: A mathematical representation of the relation between the input and output of a linear time-invariant system.

Is a compressor a linear time invariant system? I dont think so. Send two kick drum hits to a compressor thats set for a medium attack and a long decay. The output will depend on how much time is between the the two hits. Is that time invariant?
And if the output level isnt the input times a constant, which it isnt (the whole point of compressors) then its not a linear system.

 

[Cbdb2]

They will have a range in dB where they can process the signal without saturating. So, what I'm saying is, if you can line up the settings and you aren't overdriving anything, then the sound should be the same.

How many different compressors have you used? The euphonic distortion of some over driven (or too fast decay) compressors is often used. Compressors not only have attack/decay times but the curves of these can be different. Then theres feedforward and feedback designs, different dynamics. And dont forget auto attack/decay.
Ever work in a major studio?

 

[ThatSoundsGood]

So this is the kind of reasoning that I'm against. I mentioned before that the biggest gap in the profession is found in psychoacoustics. People talk about sterility and so forth but they haven't actually isolated the issue, and assume they have come to the correct conclusion without doing the hard work of testing. They talk about equipment endlessly but not about the circumstances of signal transmission and so forth, which you can only do effectively with measurements.

The testing has generally been done. The sterility that comes from Class D amplifiers in comparison to Tube amplifiers is clearly a different level of distortion and magnitude non linearities from the tube amp's inability to drive a specific impedance load. Is it not?
I'm not saying that there aren't myths in pro audio. I'm not saying that psychoacoustics are not important, but you seem to be speaking of an industry that you have little experience in. Have you worked in pro audio?

Tape saturation and compression applies only at high signal levels. If frequency response is corrected, and the signal is above the noise floor and within the speced operating range, there will be little to distinguish tape and digital media.

But we know that to achieve a useful gain structure with tape (minimize the noise floor) we have to hit the tape hard. Otherwise, it's not useful. So we prefer the compression and saturation over the noise floor and we would probably prefer it over the digital signal. Have you ever recorded to tape and to digital simultaneously and compared it?

For speakers, distortion has a much smaller role than the radiation pattern. One of main problems in audio engineering is that you can diagnose problems incorrectly, apply a fix that doesn't make sense and then be none the wiser because our hearing is so dynamic.

Distortion, radiation pattern, interaction with the room, other distortions (hysteresis, intermodulation, etc.) all play a role in how a speaker is heard. This is all easy to see in measurements and I'm quite aware of it. This is why I prefer "accurate" monitoring which includes a minimization of all of these "problems". But we cross into a different world with production when we are making choices based more on the emotion that the signal elicits.

Remember, distortion in real settings is never ever harmonic. Harmonic distortion is measured and seen only with single tone diagnostic tests. With music or any other signal is always a complex mess of nonlinearities, and many kinds of distortion are active at the same time.

Yes, this is why we minimize all of it with monitoring but control it in production. It's that simple. Accurate monitoring, tasteful production.

It's not that measurements don't help, usually the measurement is done incorrectly or there are a lack of tools. The latter is normal and is unlikely to change, particularly for accurate electrical measurements. So I think the right way to approach a situation when you dont have all the tools to make the right assessment is to resist jumping to conclusions, make careful controlled listening tests and reason from what is known of psychoacoustics.

We don't always need measurements when we know that something is completely different. It's not psychoacoustics to know that two different speakers sound different. But just because they sound different doesn't mean that we can't get both of them to sound really good to a large number of people. Also, we use a lot of measuring tools to do so.

 

[mixsit]

"Back in the day, engineers would crawl around the control room with test tones, hunting for the one spot where the bass response actually worked. When they found it, they’d drive a large nail into the floor to mark the listening position—the bigger and shinier the nail head, the better. Everything else—speakers, console, treatment—was built around that."
Got a spot off the rack here behind me. Used to check my low end there..
Home recorder flash back.. :>)

 

[JLKingsland]

Tape saturation and compression applies only at high signal levels. If frequency response is corrected, and the signal is above the noise floor and within the speced operating range, there will be little to distinguish tape and digital media.

Hey there,
I've worked in studios for 20 years and also own a pro audio manufacturer along with all of the testing accoutrements. As a matter of course we frequently do blind A-Bx testing when making changes in circuits. All that to say, I have some experience and some qualifications to speak on the above.

I appreciate the desire to have science and human sciences brought to audio. We all know that the industry can be filled with bad actors. That said, the above is a troublesome statement. Let's invent a typical setup of a standard 24 track recorder @ +3/250 nW/m ( a common calibration at this point) on +9 tape such that +3dB VU is the maximum tape level. Now, with modern (as in the last 50 plus years) close micing on drums you can easily have 15-20 dB(or more) of transient response beyond the average signal. So, even if you record your drumset at very modest levels like -6 dB VU, you will have easily 10 dB of transient clipping. Now, this isn't necessarily a bad thing if thats a sound you like. Let alone if you're using +6 tape.

Secondarily, the idea that most compressors are identical is simply false. There are a litany of factors that accumulate to determine the outcome of the sonics that are of inclusive but not limited to the gain control element. Off the top of my head they can be:
1.) Input and output interfacing, I.E. transformers vs electronic
2) Line Amps
3.) Detector circuitry
4.) Feedback methodology

Ostensibly, the methodology in which the parts are employed far outweigh the parts themselves.

Not trying to be incendiary, look forward to your reply.

 

[tengiz]

Here’s another bit of audio-adjacent folklore from 1980s. I used to know a young engineer - military encrypted comms - voice scrambling, frequency hopping, SSB modulation - that kind of thing. At some point he picked up something about subharmonics in certain oscillations - like vocal fry - and it stuck with him.

He kept going on about it until it turned into a full-on theory: every reputable audio manufacturer - consumer, pro, even military - is perfectly open about harmonic distortion. It’s right there in the specs. But subharmonic distortion? Total radio silence. Not a whisper. Not a footnote.

His conclusion? There is a cover-up. Someone, somewhere, does not want you to know what is lurking one octave below.

 

[Curvature]

Hey there,
I've worked in studios for 20 years and also own a pro audio manufacturer along with all of the testing accoutrements. As a matter of course we frequently do blind A-Bx testing when making changes in circuits. All that to say, I have some experience and some qualifications to speak on the above.

I appreciate the desire to have science and human sciences brought to audio. We all know that the industry can be filled with bad actors. That said, the above is a troublesome statement. Let's invent a typical setup of a standard 24 track recorder @ +3/250 nW/m ( a common calibration at this point) on +9 tape such that +3dB VU is the maximum tape level. Now, with modern (as in the last 50 plus years) close micing on drums you can easily have 15-20 dB(or more) of transient response beyond the average signal. So, even if you record your drumset at very modest levels like -6 dB VU, you will have easily 10 dB of transient clipping. Now, this isn't necessarily a bad thing if thats a sound you like. Let alone if you're using +6 tape.

Secondarily, the idea that most compressors are identical is simply false. There are a litany of factors that accumulate to determine the outcome of the sonics that are of inclusive but not limited to the gain control element. Off the top of my head they can be:
1.) Input and output interfacing, I.E. transformers vs electronic
2) Line Amps
3.) Detector circuitry
4.) Feedback methodology

Ostensibly, the methodology in which the parts are employed far outweigh the parts themselves.

Not trying to be incendiary, look forward to your reply.

This thread is getting somewhat circular and I am repeating myself. Apologies for not replying in detail.

I basically agree with all of your points. We can even take them a step further and say that particular combinations of gear have their own interesting interactions. In the all analog days, long, supposedly transparent chains changed the signal well before it hit the medium. Just one example of this is the dipping of the ends of frequency response as -3dB points stacked on top of each other.

My main contention is that if we understood the behaviour of these devices more precisely, then we would have more control over the result and less attachment to individual pieces of gear. It is not that all compressors (or whatever) are the same. My point about tape was that you can identify regions of linearity, however limited, and those regions are as the clean-sounding as the next thing. As such, if you continue that line of argument, I am also convinced that given a controlled set of circumstances and testing that the perceived differences of many would shrink significantly, if not disappear.

 

[Curvature]

Here’s another bit of audio-adjacent folklore from 1980s. I used to know a young engineer - military encrypted comms - voice scrambling, frequency hopping, SSB modulation - that kind of thing. At some point he picked up something about subharmonics in certain oscillations - like vocal fry - and it stuck with him.

He kept going on about it until it turned into a full-on theory: every reputable audio manufacturer - consumer, pro, even military - is perfectly open about harmonic distortion. It’s right there in the specs. But subharmonic distortion? Total radio silence. Not a whisper. Not a footnote.

His conclusion? There is a cover-up. Someone, somewhere, does not want you to know what is lurking one octave below.

I have a similar story, sort of. An engineer was recording a massive organ and was particularly impressed by the "subharmonics" he heard while monitoring over headphones. Those turned out to be, once he took them off, the low growl of the portable generator he was using to power his gear.

 

[JLKingsland]

This thread is getting somewhat circular and I am repeating myself. Apologies for not replying in detail.

I basically agree with all of your points. We can even take them a step further and say that particular combinations of gear have their own interesting interactions. In the all analog days, long, supposedly transparent chains changed the signal well before it hit the medium. Just one example of this is the dipping of the ends of frequency response as -3dB points stacked on top of each other.

My main contention is that if we understood the behaviour of these devices more precisely, then we would have more control over the result and less attachment to individual pieces of gear. It is not that all compressors (or whatever) are the same. My point about tape was that you can identify regions of linearity, however limited, and those regions are as the clean-sounding as the next thing. As such, if you continue that line of argument, I am also convinced that given a controlled set of circumstances and testing that the perceived differences of many would shrink significantly, if not disappear.

In fairness to circular conversations, this thread really is leaving science behind for hearsay and speculation. Again, not trying to be incendiary, we all do that from time to time in our own way. BUT I do think it's important to maintain accuracy on these types of forums such that science and speculation are not conflated.

Referenced to your point regarding tape: I understand what you're saying, but it's just a very flawed point. It's akin to saying, there's little percievable difference between the horse drawn carriage and a Tesla going 85 mph, if you were to implement modern braking, suspension, infotainment, steering, and climate control. Where as with tape the "linear" point your describing doesn't really exist.

We'll reinvent recorder I described above. +3/250 nW/m, AES equalization (thusly 30 ips), +9 tape. If you were to reduce the level of the input such that it wasn't do any type of the transient clipping, roughly you would trying to be peak around -17 dB VU. This would leave you with a SNR of about 40. You could use Dolby SR to get about 60 but it certainly isn't always 1:1 linear with the source. Even then you will have some degree of wow and flutter. You could use digital correction to fix it but that's not fully linear. This is all everything is working properly conditions, let alone any real world deviation, like tape quality, head quality, the machine itself both electronically an mechanically.

Short of it is, I understand the sentiment, but the tape argument is a particularly bad one. There is no concevable real world circumstance that would back up this stance.

Now if you want to talk about modern audio interfaces, that are going to basically use similar ICs with similar ADCs and DACs? Sure! That makes tons of sense. Now, circuit topology can certainly change things even with ICs, but there's only so many ways to do a differential driver and reciever.

Back to compressors? FAR too many variables. Even at the most basic layer, you do have common controls like threshold, attack, release, ratio. But there's also different sidechain designs such that one is using an timed sampling of peak averages to determine as to when threshold has been surpassed, others might use RMS, some use a sum of all these points. In that one arena alone, even if the line amp and the gain control elements were to be identical, you could have the difference of an extreme peak detector or an RMS one and the compression would NEVER line up.

Again, much respect. Thanks for the reply.

 

[Curvature]

In fairness to circular conversations, this thread really is leaving science behind for hearsay and speculation. Again, not trying to be incendiary, we all do that from time to time in our own way. BUT I do think it's important to maintain accuracy on these types of forums such that science and speculation are not conflated.

Referenced to your point regarding tape: I understand what you're saying, but it's just a very flawed point. It's akin to saying, there's little percievable difference between the horse drawn carriage and a Tesla going 85 mph, if you were to implement modern braking, suspension, infotainment, steering, and climate control. Where as with tape the "linear" point your describing doesn't really exist.

We'll reinvent recorder I described above. +3/250 nW/m, AES equalization (thusly 30 ips), +9 tape. If you were to reduce the level of the input such that it wasn't do any type of the transient clipping, roughly you would trying to be peak around -17 dB VU. This would leave you with a SNR of about 40. You could use Dolby SR to get about 60 but it certainly isn't always 1:1 linear with the source. Even then you will have some degree of wow and flutter. You could use digital correction to fix it but that's not fully linear. This is all everything is working properly conditions, let alone any real world deviation, like tape quality, head quality, the machine itself both electronically an mechanically.

Short of it is, I understand the sentiment, but the tape argument is a particularly bad one. There is no concevable real world circumstance that would back up this stance.

Now if you want to talk about modern audio interfaces, that are going to basically use similar ICs with similar ADCs and DACs? Sure! That makes tons of sense. Now, circuit topology can certainly change things even with ICs, but there's only so many ways to do a differential driver and reciever.

Back to compressors? FAR too many variables. Even at the most basic layer, you do have common controls like threshold, attack, release, ratio. But there's also different sidechain designs such that one is using an timed sampling of peak averages to determine as to when threshold has been surpassed, others might use RMS, some use a sum of all these points. In that one arena alone, even if the line amp and the gain control elements were to be identical, you could have the difference of an extreme peak detector or an RMS one and the compression would NEVER line up.

Again, much respect. Thanks for the reply.

I honestly think our difference in perspective is not that wide. Yes, fully agree that analog media show significant deviations in real (not contrived) use even with studio-grade Studers and the like, not to get started on vinyl. And the point has been made every few posts, and I am fully aware, of how different compressor operations are. I don't know of good research that demonstrated how we should interpret the difference between each one, though. That they are different is without question: how much of that difference matters? It's well-known that distortion and differences are more easily measurable than audible. And, separately, even if the behavior of the gear is very different, does their real-world use in studios make use of that difference or do engineers consciously or unconsciously try for convergence? There has certainly been a lot of work on comparing recording media, since the goal is clear (preservation of the signal) and there are many tools that can effectively measure deviations. That not only makes the scientific work easier, it also gives a firm quantitative foundation for controlled listening tests. The same can't be said for the tools of production for exactly the reason you noted: too many variables and unclear targets.

I don't know if this point was ever said explicitly, but: folklore makes the job of the engineer harder. The goal as far as I see it is to give good guidance. The differences between gear are just the starting point, IMO.

Here's another story in that vein, sort of. Yehudi Menuhin and Glenn Gould, in an interview, were discussing recording technique. Glenn Gould popularized the use, in classical piano, of putting together the final recording from multiple takes. There was some idea before that the integrity of the recording and its moment had to be preserved. Not only that, there was some sense in the prevailing musical philosophy that you could not simply repeat middle passages without recapitulating what led up to them to ensure continuity of intent. That put a lot pressure on the performer to get the take perfect in one shot, without musical mistakes, and pressure on the engineer to get the take perfect from a recording perspective.

 

[dfuller]

Tape saturation and compression applies only at high signal levels. If frequency response is corrected, and the signal is above the noise floor and within the speced operating range, there will be little to distinguish tape and digital media.

Yeah, but tracking to tape was almost always done as hot as you could get away with in order to improve the snr.

 

[JLKingsland]

I honestly think our difference in perspective is not that wide.

I don't disagree. As spelled out above, I think it's important to speak on topics like this with nuance.

And the point has been made every few posts, and I am fully aware, of how different compressor operations are. I don't know of good research that demonstrated how we should interpret the difference between each one, though. That they are different is without question: how much of that difference matters?

Generally a product spec sheet tells you quite a bit of about the nature of what a compressor can do. A LA2A has a .01 second attack time, a Spectra Sonics 610 has a 0.0000001 second attack time. What this imminently tells you is that the spectra sonics compressor is suitable for managing transients in a way that the LA-2a simply will not. This is not inaudible. How much the difference matters in these 2 pieces depends on what you're trying to do.

If you want to control the transient response of a drums set? Incredibly meaningful. The LA2A simply isn't fast enough to accomplish this.

Level out a few dB of dynamic range on a bass? Probably less meaningful. BUT the release time of the LA2A isn't linear(60 ms for the first 50 percent of release, and 500 ms to 15 seconds for the other 50 percent dependant on program) so that will change aspects as to how the bass sounds, and debately more important how it sits and is presented in a mix. Again very different.

I have a question for you. I appears at time that lines of question, such as "How much of that difference matters?", is an attempt to make all things the same, or functionally the same, or close enough to the same. There seems to be at times borderline bias toward differences not mattering. So, what is the headspace and reasoning behind those lines of questioning?

And, separately, even if the behavior of the gear is very different, does their real-world use in studios make use of that difference or do engineers consciously or unconsciously try for convergence? There has certainly been a lot of work on comparing recording media, since the goal is clear (preservation of the signal) and there are many tools that can effectively measure deviations. That not only makes the scientific work easier, it also gives a firm quantitative foundation for controlled listening tests. The same can't be said for the tools of production for exactly the reason you noted: too many variables and unclear targets.

This is a question to be addressed on a per engineer per source basis. In my case, yes, I do take advantage of these differences.

Here's another story in that vein, sort of. Yehudi Menuhin and Glenn Gould, in an interview, were discussing recording technique. Glenn Gould popularized the use, in classical piano, of putting together the final recording from multiple takes. There was some idea before that the integrity of the recording and its moment had to be preserved. Not only that, there was some sense in the prevailing musical philosophy that you could not simply repeat middle passages without recapitulating what led up to them to ensure continuity of intent. That put a lot pressure on the performer to get the take perfect in one shot, without musical mistakes, and pressure on the engineer to get the take perfect from a recording perspective.

This is the interesting part, the feedback loop between process and performance. Equally referenced things about, say compressors, compression can change the way someone performs. IF it's set up in such a way that it feels "good" to a performer, there's a reasonable chance their performance may be different. This definitely gets in to the human sciences or anecdotal suggestion end of the pool. Not that that's a bad thing, but you certainly can't put an analyzer on that.

 

[tengiz]

Here's another story in that vein, sort of. Yehudi Menuhin and Glenn Gould, in an interview, were discussing recording technique. Glenn Gould popularized the use, in classical piano, of putting together the final recording from multiple takes. There was some idea before that the integrity of the recording and its moment had to be preserved. Not only that, there was some sense in the prevailing musical philosophy that you could not simply repeat middle passages without recapitulating what led up to them to ensure continuity of intent. That put a lot pressure on the performer to get the take perfect in one shot, without musical mistakes, and pressure on the engineer to get the take perfect from a recording perspective.

Horror! What’s next - auto-tuning Yehudi Menuhin, quietly slipping it into the catalog, and then assuring us with a straight face that "nothing essential was changed"?