The Fight for FM

[ouroboros]

I have always wondered if their are any accessible academic discussions about how people hear this kind of thing.

Yes, it's called synesthesia and it's an interesting medical condition that's apparently very frequently found with electronic musicians.

http://en.wikipedia.org/wiki/Synesthesia

[edit] or more specifically:

http://en.wikipedia.org/wiki/Synesthesi ... ynesthesia

[Aroused by JarJar]

I have always wondered if their are any accessible academic discussions about how people hear this kind of thing.

Yes, it's called synesthesia and it's an interesting medical condition that's apparently very frequently found with electronic musicians.

http://en.wikipedia.org/wiki/Synesthesia

[edit] or more specifically:

http://en.wikipedia.org/wiki/Synesthesi ... ynesthesia

Certainly, that's the subjective part of it, though there is a spooky amount of shared subjectivity, I'll get to those things later (already touched upon it earlier).

But it is very important to understand that the first, historical (ancient) use of the word "color" in reference to sound means the actual physical spectra. That is only subjective on a minimal level (I probably can't hear the extreme highs like a teenage girl could for example).

When I say "spooky", it's not really spooky if you really think about it, and are aware of what's going on on a physical, objective level, as well as putting things in social and historical perspectives.

My son hadn't even been speaking very long when he asked, listening to a solo violin, "Why is the lady so sad?" He'd never heard the cliche "weeping violins", but there are physical, objective reasons, and human, animal, and cross-cultural, reasons, why a solo violin really does sometimes sound like a weeping lady to a whole lot of human beings.

@ dan_s:

To me, your example doesn't sound "orange", it sounds like your typical low-quality (like, oversampling and hope for the best rather than running envelopes at audio rate and so on because users will whine about CPU usage kind of thing) software "FM" rammed through a bit-crusher.

[z15]

The funny thing is that the "deniers" here have no problem accepting that apparently "orange" or "brown" is perceived more or less the same by everyone (it is, there have been studies down using functional MRI scans), but won't accept anything similar for the perception of audio.

The problem there is that perception of colours has been widely studied and measured. And even then, there are various conditions where relative brightness of two colours may appear different in different lighting conditions, there are a lot of optical illusions that show that same colour can be perceived different by one person (the checker shadow illusion), or two different colours the same. Et cetera. Also witness that human visual system adapts to different conditions, while something like a camera doesn't, unless a model of adaptation is programmed to it somehow. And mind you, we're speaking about clear concepts like "which one of these greys is brighter" or "is this colour red or brown" that have some reasonably well-defined meaning.

Well, human hearing adapts to different things too, there is masking going on, et cetera. Human perception isn't as simple as "there are odd harmonic partials so the sound must be metallic". With audio, we don't even have too much scientific data about how mood, timbre, et cetera is perceived - those are higher level features that we have no direct mapping to low level physical features for.

Jar Jar gives a good point I agree with: you can define a "clean FM algorithm" quite objectively. We can say that a timbre of a clarinet can approximately be described by such and such features by measuring sound. However, subjective adjectives are not necessarily something even a big group of people agree on, no matter how many people one knows that agree with his taste. If ten thousand hobbyist musicians agree that FM8 sounds "warm" and a thousand engineers agree that it sounds "cold", does that mean that the cultural consensus is that the overall sound is "cold" because the ten thousand obviously just have bad ears? Again, probably the 11000 people would all agree that a red carpet is a red carpet.

Even a manual human-made test "what mood does this song have" inside one culture doesn't produce even near unanimous response. And even if these things are found to be agreed by majority of people, we still need to figure out the mapping to certain properties of the sound which is nowhere near trivial in practically any case. As long as we don't even have clear proof that some subjective qualities are agreed by majority of people, we can't really say that "this subjective quality means that these features in signal are present" - it would be misleading to everyone except the group of people who agree.

Think about an adjective such as "clean". I consider my bedroom / home studio quite clean at the moment, but someone with a different view to "clean" might as well say it's a f**king mess. If the room was empty and all the surfaces were cleaned up, most people would agree it's "clean". However, add stuff to the picture and the amount and organization of the said stuff gives different people different perceptions of how "clean" the room is. However, practically all of those people would notice that the doors of my clothes drawer are white, and that the window panes are sort of, um, brownish dark red? Oops, even that might be a point of debate.

[nuffink]

The funny thing is that the "deniers" here have no problem accepting that apparently "orange" or "brown" is perceived more or less the same by everyone (it is, there have been studies down using functional MRI scans), but won't accept anything similar for the perception of audio.

I will. The minute you show me the results of the MRI scans done on the heads of MEN WHO LISTEN TO D/A CONVERTERS indicating that they all hear a Euphonium as ginger with red spots. Or something similar.

I'm quite keen on evidence. I rarely argue against it. Got any?

[TristezaOrange]

they all hear a Euphonium as ginger with red spots

Also, I think I'm going to start a band or a collaborative project here in KvR called Men Who Listen To D/A Converters .

[z15]

But it is very important to understand that the first, historical (ancient) use of the word "color" in reference to sound means the actual physical spectra. That is only subjective on a minimal level (I probably can't hear the extreme highs like a teenage girl could for example).

The problem is kind of that psychoacoustics come to play here, which makes it difficult to have a clear mapping to physical properties.

For example, think about the missing fundamental - the pitch is perceived not only by the actual frequencies present but also by the relative strength of the overtone series. So, for example, if you have a harmonic series of 200-300-400-500-600 Hz in the right proportions, you may perceive that the sound also has the 100Hz component and think "wow, this is phat, there's some serious bass going on". That's how some "bass enhancement" plugins and devices work - you aren't actually increasing the low frequency content but adjusting certain upper harmonics, and the end result is a sensation of more bass.

Another example are the psychoacoustic enhancers / exciters. You can for example remove / attenuate the original high frequency content and resynthesize it by distorting some band of the signal, or you can adjust the perceived amount of bass by altering the phase structure of the signal. Again, hard to define what the exact mapping from "this sounds more bassy" or "this sounds more bright" to a change in frequency content would be in general, since it isn't just a change like "this particular frequency range is boosted to make it brighter / bassier".

That's psychoacoustics, and we haven't even eliminated the psychological factor such as associating how something looks to how it sounds, or associating a certain adjective because someone else told you so, even though you had no association like that before and many others might still not have it. Or your ears having gotten used to something (monitoring setup, certain instruments you use) and perceiving differences of the spectra differently than someone else.

My son hadn't even been speaking very long when he asked, listening to a solo violin, "Why is the lady so sad?" He'd never heard the cliche "weeping violins", but there are physical, objective reasons, and human, animal, and cross-cultural, reasons, why a solo violin really does sometimes sound like a weeping lady to a whole lot of human beings.

Yeah, that is because the sound is so similar to another sound and humans are good at associating things. Sometimes when the associations are straight enough and both parts of them belong to a certain culture, a very large group of people agree with them - and sometimes when they are more "far-fetched", it's kind of dangerous to assume that IMHO.

The "metallic" example - it's reasonable to assume that many people have heard clangs that different pieces of metal make, and thus often agree that something sounds metallic. Provided that the metallic objects associated are same kind (think hitting two forks together, vs banging a huge metal door, vs a tuning fork), it's probable that a large group of people agree. Less when the association is to some other sense, such as "cold/warm" or "blue". It's easy to imitate and analyze a group of "metallic" sounds by sampling hits of metal (no, not the '80s ones), doing frequency analysis and so on, but to sample a "blue" sound we first need to agree what a "blue" sound is like - obviously it's not playing a blue guitar, or hitting a blue piece of metal to another blue piece of metal. Nor is "warm" sound about sampling a warm meal or an oven.

I guess my point is that if you compare something to something else that makes a sound and the adjective is related to that particular object, it's much easier to agree than when the adjective isn't related to the object itself. And even when talking about associating two different sounds to have the same quality, it most often isn't trivial to measure what the particular quality is.

FWIW, you have quite a lot of good points and aren't spewing bullcrap by no means, I just want to argue with you since I don't agree with all of the points

[whyterabbyt]

Of course, even if there was a common consensus as to whether a sound was 'orange', the problem is that that goes right out the window when it becomes a question of being 'more orange' than another sound.

in fact, is there even a consensus (proven by MRI or not) that a given thing might be 'more orange' in colour than another not-quite-as-orange thing? or isnt that the point at which we have to delve into adjectives and modifiers.

beyond that, the idea of 'most orange' doesnt even exist, let alone as a consensus, and yet that's what we're seemingly getting passed off to us as definitive, no less, when it comes to synths and audio.

[Aroused by JarJar]

The funny thing is that the "deniers" here have no problem accepting that apparently "orange" or "brown" is perceived more or less the same by everyone (it is, there have been studies down using functional MRI scans), but won't accept anything similar for the perception of audio.

The problem there is that perception of colours has been widely studied and measured. And even then, there are various conditions where relative brightness of two colours may appear different in different lighting conditions, there are a lot of optical illusions that show that same colour can be perceived different by one person (the checker shadow illusion), or two different colours the same. Et cetera. Also witness that human visual system adapts to different conditions, while something like a camera doesn't, unless a model of adaptation is programmed to it somehow. And mind you, we're speaking about clear concepts like "which one of these greys is brighter" or "is this colour red or brown" that have some reasonably well-defined meaning.

Well, human hearing adapts to different things too, there is masking going on, et cetera. Human perception isn't as simple as "there are odd harmonic partials so the sound must be metallic". With audio, we don't even have too much scientific data about how mood, timbre, et cetera is perceived - those are higher level features that we have no direct mapping to low level physical features for.

Jar Jar gives a good point I agree with: you can define a "clean FM algorithm" quite objectively. We can say that a timbre of a clarinet can approximately be described by such and such features by measuring sound. However, subjective adjectives are not necessarily something even a big group of people agree on, no matter how many people one knows that agree with his taste. If ten thousand hobbyist musicians agree that FM8 sounds "warm" and a thousand engineers agree that it sounds "cold", does that mean that the cultural consensus is that the overall sound is "cold" because the ten thousand obviously just have bad ears? Again, probably the 11000 people would all agree that a red carpet is a red carpet.

Even a manual human-made test "what mood does this song have" inside one culture doesn't produce even near unanimous response. And even if these things are found to be agreed by majority of people, we still need to figure out the mapping to certain properties of the sound which is nowhere near trivial in practically any case. As long as we don't even have clear proof that some subjective qualities are agreed by majority of people, we can't really say that "this subjective quality means that these features in signal are present" - it would be misleading to everyone except the group of people who agree.

Think about an adjective such as "clean". I consider my bedroom / home studio quite clean at the moment, but someone with a different view to "clean" might as well say it's a f**king mess. If the room was empty and all the surfaces were cleaned up, most people would agree it's "clean". However, add stuff to the picture and the amount and organization of the said stuff gives different people different perceptions of how "clean" the room is. However, practically all of those people would notice that the doors of my clothes drawer are white, and that the window panes are sort of, um, brownish dark red? Oops, even that might be a point of debate.

Yes, it's a matter of deciding where to draw the (fuzzy) lines.

[z15]

Of course, even if there was a common consensus as to whether a sound was 'orange', the problem is that that goes right out the window when it becomes a question of being 'more orange' than another sound.

This was something I was also going to write about but forgot while rambling as usual. I guess that in theory you COULD define something to be "as red as possible" by aiming to produce only the associated wavelength / colour component as accurately as possible. In the same manner, you could make a sound "pure and clean" by having a sinewave that is reproduced with as little distortion as possible. Those are the trivial cases.

However, the farther away you get from the simple cases, as you start mixing colours / tones, and actually approaching the subject from the "how humans perceive differences" angle instead of "what do the measurements say" angle, the harder it is to pinpoint relative differences such as that - in the sense that some quality described by an adjective changes and you are able to clearly specify or measure a difference in the signal. If you take an adjective such as "clean" and use it with music, it's hard to specify "cleaner" or "cleanest" by other means than how small amount of harmonics and noise it has - the ideal being a pure noiseless sinewave. For example. You can make "clean" context-dependent but it falls into realm of subjectivity pretty fast unless you are talking about the performance of an algorithm compared to the ideal one and not about how you perceive sound. Sometimes they may correlate, sometimes not.

That doesn't mean that subjective descriptions are useless - they're an invaluable form of communicating with other people. It's just that claiming some subjective adjective is shared by some particular group of people who I know isn't enough to convince anyone that it's clear and definable what that adjective means, or what more of that particular quality would mean in a given case. It's bit like talking with your friends about music genres and claiming "Black Sabbath is more rock than Led Zeppelin, practically everyone I know agrees with that and they have listened to a lot of rock".

[z15]

Yes, it's a matter of deciding where to draw the (fuzzy) lines.

True, I guess my point was just that everyone draws their lines themselves. I think you said earlier about using "fuzzier" or more subjective descriptions, and trying to provide examples of those as "this is what I mean by this adjective". That's a good way to go, especially if you're able to analyze carefully what is in common with examples given.

However, I think it's kind of dangerous to draw conclusions like "well, it should be obvious what this adjective means, plenty of people agree that it means these kinds of sounds or qualities". Or to claim that the qualities described by the adjectives would be simply a matter of mapping them to the spectrum of a sound in some understandable way. The background knowledge isn't same for all of the people, nor is taste or preconceptions, and the human perception is a lot weirder than just taking a fourier transform of the spectrum and looking at it.

[Aroused by JarJar]

Yes, it's a matter of deciding where to draw the (fuzzy) lines.

True, I guess my point was just that everyone draws their lines themselves. I think you said earlier about using "fuzzier" or more subjective descriptions, and trying to provide examples of those as "this is what I mean by this adjective". That's a good way to go, especially if you're able to analyze carefully what is in common with examples given.

However, I think it's kind of dangerous to draw conclusions like "well, it should be obvious what this adjective means, plenty of people agree that it means these kinds of sounds or qualities". Or to claim that the qualities described by the adjectives would be simply a matter of mapping them to the spectrum of a sound in some understandable way. The background knowledge isn't same for all of the people, nor is taste or preconceptions, and the human perception is a lot weirder than just taking a fourier transform of the spectrum and looking at it.

I think it's also important not to let yourself get bullshitted into a helpless position of complete vagueness and relativity. Draw your (fuzzy) lines, as an individual, salting always with the brutal (or comforting?) fact that no person is an island. IMHO.

Well, there's no choice but to do exactly this in the professional world, unless you want to wind up in the madhouse (which I hear isn't so bad, but that's from a friend who literally came running down the mountain naked from the aforesaid place, :LOL: quite a good artist, too)

The "definitive statements" I've noticed on this thread are objective measurable ones such as my own (let's reckon things like rise times and general lack of calculation error, and variation over time (aka motion), shall we, it's not rocket science) and, well, "homeless person duking it out with their demons" style statements of esoteric personal perception parading as "universal truths" such as: "there is no such thing as fundamental sound quality".

Which was a mind-boggling statement, poetry even: for it states that, for example, audible ground hum on a hardware synthesizer has no part in the physical spectra of the sounds produced by that synthesizer, and therefore no bearing on our subjective perceptions of those sounds.

Whoo-hoo! I must confess that I'm quite pleased by the exotic nature of such bizarre claims because the social structure where I live is such that there are almost no rambling homeless people around, and to be frank- I miss them.

The background knowledge isn't same for all of the people, nor is taste or preconceptions, and the human perception is a lot weirder than just taking a fourier transform of the spectrum and looking at it.

Certainly. Obviously we do what we can with what we have, in the time we have.

Edit- anyway, we've got to get back to the FM sounds!

Still would like to hear a hardware version of the simple "china" kind of sound we had going earlier.

[dan_s]

@ dan_s:

To me, your example doesn't sound "orange", it sounds like your typical low-quality (like, oversampling and hope for the best rather than running envelopes at audio rate and so on because users will whine about CPU usage kind of thing) software "FM" rammed through a bit-crusher.

the orange part was a joke. It is a hardware dx.

point being, it was recorded through a $30 mixer.

[nuffink]

The "definitive statements" I've noticed on this thread are objective measurable ones such as my own (let's reckon things like rise times and general lack of calculation error, and variation over time (aka motion), shall we, it's not rocket science) and, well, "homeless person duking it out with their demons" style statements of esoteric personal perception parading as "universal truths" such as: "there is no such thing as fundamental sound quality".

Which was a mind-boggling statement, poetry even: for it states that, for example, audible ground hum on a hardware synthesizer has no part in the physical spectra of the sounds produced by that synthesizer, and therefore no bearing on our subjective perceptions of those sounds.

Ten out ten for chutzpah.

It's you making the extraordinary claims. The very idea that there is such a thing as fundamental sound quality is an extraordinary claim an a such requires extraordinary evidence. Show us the evidence.

edit: This is where you deny that you were talking about fundamental sound quality as a universal value judgement and reach for the cSound manual again.

[Aroused by JarJar]

The "definitive statements" I've noticed on this thread are objective measurable ones such as my own (let's reckon things like rise times and general lack of calculation error, and variation over time (aka motion), shall we, it's not rocket science) and, well, "homeless person duking it out with their demons" style statements of esoteric personal perception parading as "universal truths" such as: "there is no such thing as fundamental sound quality".

Which was a mind-boggling statement, poetry even: for it states that, for example, audible ground hum on a hardware synthesizer has no part in the physical spectra of the sounds produced by that synthesizer, and therefore no bearing on our subjective perceptions of those sounds.

Ten out ten for chutzpah.

It's you making the extraordinary claims. The very idea that there is such a thing as as fundamental sound quality is an extraordinary claim an a such requires extraordinary evidence. Show us the evidence.

You would claim that for example a big ass mains hum in the sound of hardware synth is not something we could call part of the fundamental sound quality of that synthesizer?

Screw my 10 for 10, you get 10 squared of 10 for bravado and derring-do beyond the call of the gojim, mensch. Mel Gibson eat your heart out!

[nuffink]

Don't give me that bollocks. You're trying to slide out from a position that even you can see is untenable. If there was universal agreement on sound quality the only deciding factor in equipment choice would be price.

Back down with a bit of grace. You're flogging a dead horse and you know it.