Our perception of audio

[Kal Rubinson]

Have you tried listening to natural sound recordings, like birds in a soundscape? You may be able to judge the realism, or the plausibilty at least, of what you hear? Even if you were not at the recording session?

In theory, that is interesting but I do not think it will really work. I have not studied nor am I able to distinguish among bird voices in any meaningful way so how could I "judge the realism, or the plausibilty" of what I hear?

 

[Ron Texas]

I hear, therefor I am.

 

[Theo]

Well, I did. I also recorded some sounds myself. It is actually very easy to perceive the geometry of the recorded events when listening to a good system. Being able to check on the veracity of the bird song itself is another story.... As it would rely on perception memory, It probably wouldn't mean anything.
And of course, there is no way that you would be able to quantify the parameters of the system just by listening to it....

 

[NorthSky]

I too recorded sounds of nature, and live music too, in the part, like millions of nature lovers and music mixing regarding emgineers. Playing back some of those tapes with water falling from waterfalls, rain, trains passing by, winds relaxing and zigzagging between top and trees and branches of the forests, eagles clapping wings, echoes reflecting the surface of lakes, fishes jumping, salmons against the current, ... it can be very spooky in a a good multichannel setup. It's almost ghosting, fooling, revelator.
Our brain fills the missing dots, we hear music in the background coming from the cabin by the lake with the musicians inside. The winds transport it through the woods, reverberating from the bark of the trees and reflecting though the water surface, and echoing across the lake on the next forest.

The way our brain perceive sounds from the woods (audio) is the way our quality live recordings reflect our passion and the way we play all it back with that same passion, dedication, attention.

 

[Nowhk]

Some of all of that. The way I see it is this:

What reaches our ears is 'a signal', a 2D wiggly line, but our brains decode it into the original acoustic 'objects', and at a higher level they also follow progressions of rhythm and melody, etc.

The brain could be mistaken about the acoustic objects: what sounds like two people singing could really be four, but one of the singers might be singing in exact antiphase to one of the other singers and cancelling them both out at the microphone - but this is very unlikely! Maybe the vocalists are adept at singing phonemes in clever rotation and chopping the words up between them. Anything is possible, but in reality, our brains use logic, familiarity and continuity to decode the signal back into the most likely integral objects.

Each object has its own timbre, rather than 'the signal' having timbre.

Without knowing about the nature of microphones, cables, recording systems, speakers, human listeners would never think in terms of 'the signal', but would only think in terms of the objects. An orchestra conductor would never say "I need more mid range!"; he would say "I need more violins! And less of the trumpets". He wouldn't say to the whole orchestra "I don't like your timbre...", but he might say "The trombone is too strident...".

But poor reproduction (frequency response, distortion, phase, timing) means that in audio, the acoustic objects are often blurred together in *meaningless* mechanical/electronic-related ways that have nothing to do with real sound and acoustics. And this, coupled with knowledge about the mechanics of audio reproduction, is where the audiophile transition is made from objects to 'signal'.

Once the separation between the objects is degraded sufficiently, all that is left is to adjust the 'colour', 'flavour' or 'timbre' of the stream of lumpy audio paste - and people can actually enjoy listening to it, but it is a different mode of listening compared to musical 'objects' in the live situation. The higher level stuff (melody etc.) probably remains intact, but some of the lower level complexity and/or simplicity is gone, replaced with a synthetic, artificial, uniform substitute.

As this substitute is regarded as 'high end' there is no reason for anyone to strive for anything better.

I'm still not quite sure what do you mean with these "audio/musical/acoustic objects". What are they?
When listening to music what I have are "percepts"; object become in the domain of "think, reasoning and imagination" (abstract objects?).

No? Is there any example/literature about those "objects"? Really can't sort it out...

Have you tried listening to natural sound recordings, like birds in a soundscape? You may be able to judge the realism, or the plausibilty at least, of what you hear? Even if you were not at the recording session?

That's another thing. Its not "playing back" somethings: its by definition a single/unique experience. If I go to the park twice, I'll get different realism, sounds, perceptions, and so on. Its unique and unrepeatable.

My doubts become from the fact that it seems that also "recordings", after all, its somethings similar. But with the initial (illusory) intent to preserve somethings "fixed", when in fact this won't never happens.

 

[Cosmik]

I'm still not quite sure what do you mean with these "audio/musical/acoustic objects". What are they?
When listening to music what I have are "percepts"; object become in the domain of "think, reasoning and imagination" (abstract objects?).

For example, the objects could be:

• a guitar and a singer
• a singer and an audience
• a singer and a church
• a sad voice and a happy voice
• a woodwind section and a string section
• a lone voice in a silent room

Or perhaps it could be:

• a melody and a harmony
• a melody and percussion backing
• a sweet harmony and an angry voice
• a modern sound and an old-fashioned melody
• an increase in tension and a resolution
• a beautiful memory and an optimistic future
• the planets of the solar system
• nostalgia
• fear of the unknown

At any one time our brains might sort what it is hearing into such 'objects', and possibly into a different set of allocations of objects the next time it hears the same recording. The point being that 'the signal' is not a monolithic object in itself, yet audiophiles treat it as though it is. The existence of the concept of 'mid range' or wanting to apply 'a modicum of second harmonic distortion to the signal' gives the game away.

'The signal' is just a by-product of recording and playing back acoustic events; it is not actually 'a thing' at all. 'It' cannot meaningfully be manipulated either deliberately or accidentally. Doing so just blurs the separation of the 'objects' in meaningless ways and reduces the number of permutations of those objects that our brains might conjure with.

Lack of fidelity results in a reduced number of 'objects'.

 

[Cosmik]

Why should our perception of audio be any different from reading a book? If we read the same book more than once we will perceive it differently from any other time we read it, but it doesn't mean that the accuracy of 'reproduction' therefore doesn't matter.

If we start to throw in a few spelling mistakes or mix up the order of the words we begin to degrade the experience of reading the book. Something new might emerge: a pleasing pattern of letters on the page or a comic aspect to the randomness of words and bad spelling, but those are just accidental and meaningless. The hobby of book randomizing and letter patterning would be a pretty shallow one.

I think that those people whose hobby is messing about with 'the signal' by deliberately adding distortion or knowingly using archaic methods are on that continuum.

 

[Blumlein 88]

Why should our perception of audio be any different from reading a book? If we read the same book more than once we will perceive it differently from any other time we read it, but it doesn't mean that the accuracy of 'reproduction' therefore doesn't matter.

If we start to throw in a few spelling mistakes or mix up the order of the words we begin to degrade the experience of reading the book. Something new might emerge: a pleasing pattern of letters on the page or a comic aspect to the randomness of words and bad spelling, but those are just accidental and meaningless. The hobby of book randomizing and letter patterning would be a pretty shallow one.

I think that those people whose hobby is messing about with 'the signal' by deliberately adding distortion or knowingly using archaic methods are on that continuum.

That is how I felt when my high school teacher insisted I read Chaucer in the middle English. It degraded the reading experience to me, but she kept insisting it was better.

When she wanted us to read Beowulf in the old English the experience was degraded so much it wasn't possible to enjoy the story. I only gained an appreciation for Beowulf when the following summer I read a translation of it into modern English.

But is reading a modern translation going away from accuracy to make it more palatable? We aren't talking about word or book randomizing. The change from one to the other had a definite non-random pattern.

Picking a non-accurate reproduction doesn't automatically equate to randomizing.

 

[Cosmik]

Picking a non-accurate reproduction doesn't automatically equate to randomizing.

But an automatic process does - or at least to arbitrariness. The automatic process has no knowledge of what is stored in 'the signal', unlike the human who translates a book into modern English. A human translating a book is the equivalent to a completely different performance of the piece, with some changes to the composition, too.

If an enthusiast wanted to break a recording of some ensemble into its component parts, for example, and then eliminate, or boost the level of, or re-tune just one of those parts, then that would be something like the above example: human intelligence being applied to a problem with prior investigation of what the signal comprises and a conscious, meaningful (possibly) translation of the recording. This is quite different from just some box designed to distort/compress/EQ 'the signal'.

 

[Blumlein 88]

If prior investigation indicates a certain distortion profile and dynamic treatment is "easy on the human ear" it isn't random. In which case it can be automatic, and added knowledge isn't needed for it to be effective.

 

[Cosmik]

If prior investigation indicates a certain distortion profile and dynamic treatment is "easy on the human ear" it isn't random. In which case it can be automatic, and added knowledge isn't needed for it to be effective.

Like 'replacing long words with shorter ones' would make a book easier to read? Sure. But in doing so the possible permutations available for the mind to conjure with are reduced.

 

[Thomas savage]

Like 'replacing long words with shorter ones' would make a book easier to read? Sure. But in doing so the possible permutations available for the mind to conjure with are reduced.

Just add pictures, I only read books with LOTS of pictures.

 

[Blumlein 88]

Like 'replacing long words with shorter ones' would make a book easier to read? Sure. But in doing so the possible permutations available for the mind to conjure with are reduced.

Shorter words are all a matter of taste and or purpose.

Easier on the ear is what some people want even if it reduces potential permutations and simplifies subtle concepts. Same for some books. For some audiences you need simple versions of both books and music.

Moving the goalposts if you insist simpler and less complex must be worse. Depends on other factors.

 

[Cosmik]

Shorter words are all a matter of taste and or purpose.

Easier on the ear is what some people want even if it reduces potential permutations and simplifies subtle concepts. Same for some books. For some audiences you need simple versions of both books and music.

Moving the goalposts if you insist simpler and less complex must be worse. Depends on other factors.

Not moving the goalposts. You are claiming that an automatic process can modify a book and make it easier on the brain without any other side effects. But of course it is throwing away information created by a living, breathing, creative human and substituting information 'created' by some algorithm (that was in turn created by a corporate drone or spotty teenager). Information is lost, meaning is distorted, and the algorithm can get it jarringly wrong.

If the process is not automatic but requires human intervention then that is like a completely new 'performance' and re-interpretation of the piece.

 

[andreasmaaan]

Why couldn’t an algorithm conceivably modify a book such that it is subjectively better?

 

[Cosmik]

Why couldn’t an algorithm conceivably modify a book such that it is subjectively better?

It could - but it would be 'better' in the sense of superficial gloss, or unintentional humour, or other spurious meaninglessness. Bad translation *could* make a dull text book on economics a viral hit for teenagers on the internet - for about a day.

 

[Cosmik]

"I've developed an algorithm to improve books"
"I've developed an algorithm to improve recordings"
To me, these statements are equivalent in terms of their credibility.

 

[andreasmaaan]

I’m sure AI or algorithm based editing of all kinds of written texts - eventually perhaps even the writing of them - will in fact be common (and best) practice in the near future.

 

[Blumlein 88]

Think of the Turing machine of Alan Turing. If a human expert can translate Beowulf and make it better for me, a modern English speaker, then some series of computer algorithmic steps at least in theory could do so as well.

 

[Blumlein 88]

"I've developed an algorithm to improve books"
"I've developed an algorithm to improve recordings"
To me, these statements are equivalent in terms of their credibility.

I agree.