Composers' Forum

Music Composers Unite!

That's idea.  What I am looking for are some NUTS & BOLTS ideas hear.

My first precept is to analyze the oscillation patterns of a recorded human voice then transpose the analysis into notation.

i.e. 10,000 violins,each given a note (or notes) to play, in a row, but all 10,000 violins would only take as long as it does to say,  "How the hey there do I do this?"

I have been milling over this idea for 20 years,

Any body out there got some good solid ideas on how to achieve it.  I am not looking for ethereal or philosophical exposition here.  What I am seeking is some down to the ground implementation tools.

 

Got a HO?

Views: 322

Reply to This

Replies to This Discussion

If pitch were the only variable, this still might never be possible, because it involves changes both extreme and nearly imperceptible, occurring nearly simultaneously.  Occurring in a manner that fingers on a fretboard would unlikely be able to reproduce.

But we're not discussing only pitch.  Complex resonance variables--again, nearly simultaneously extreme and minute--are at play.  The shape of the violin body does not change (in a significant way), while the shape interior of the mouth (and sinus) is in constant flux.  Articulation is another important concern.  I can't conceive of how even 10,000 violins could create an explosive "B" articulation, though I concede it may be possible.  There are many vocal sounds that produce no pitch, per se: "S" or "Th," for example.

Some of the world's greatest scientific minds worked for more than 150 years to recreate the human voice using technology, and their best efforts were often "barely intelligible."  More recent computer technology has allowed for much progress in this area, but most of the highest quality (to listeners) systems don't involve synthesis "from scratch," particularly when results are needed in real time (i.e., not hundreds of man-hours to create minutes of realistic recorded speech), and instead involve piecing together snippets of audio on the fly.

I suggest we get the National Weather Service working on this. They have been producing high-quality human voice simulations for deacdes now, such as:

"thenationalweatherservicveingraymainehasdetectedalineofseverethunderstormsmovingsouthat45milesperhourthesestormsareassociatedwithariskoftornadoesandseveredownpourspeopleinlowlyingareasareadvisedto takeshelterimmediately"

Wesley, for this endeavor you will need to procure an Acme frequency analyzer.

Then , simply match the waves of the 2 sources. The last one to attempt anything

like this , I believe, was Wylie Coyote back in the 1960's. But as usual, he wasn't

successful.    " Beep- Beep"

David, Thank you for your keen insight, and intelligence.  This is very very helpful.  Addressing the CHANGING shape of the vocal chords with the tongue and exhalation and mouth shape,  I would guess using various instruments would be necessary, Such as for the pronunciation of "BEE" using a tympany or Indian Drum simultaneously with a trumpet high note. I do realize the problem in the thesis is startlingly complex, I hope you are intrigued by it enough to give a few more  ideas.  Thanks,  Wesley

David J. Downs said:

If pitch were the only variable, this still might never be possible, because it involves changes both extreme and nearly imperceptible, occurring nearly simultaneously.  Occurring in a manner that fingers on a fretboard would unlikely be able to reproduce.

But we're not discussing only pitch.  Complex resonance variables--again, nearly simultaneously extreme and minute--are at play.  The shape of the violin body does not change (in a significant way), while the shape interior of the mouth (and sinus) is in constant flux.  Articulation is another important concern.  I can't conceive of how even 10,000 violins could create an explosive "B" articulation, though I concede it may be possible.  There are many vocal sounds that produce no pitch, per se: "S" or "Th," for example.

Some of the world's greatest scientific minds worked for more than 150 years to recreate the human voice using technology, and their best efforts were often "barely intelligible."  More recent computer technology has allowed for much progress in this area, but most of the highest quality (to listeners) systems don't involve synthesis "from scratch," particularly when results are needed in real time (i.e., not hundreds of man-hours to create minutes of realistic recorded speech), and instead involve piecing together snippets of audio on the fly.

Yes, But its the only crazy idea I have today.  And a real challenging one. And, I believe it possible.

Meee meeep.

roger stancill said:

Wesley, for this endeavor you will need to procure an Acme frequency analyzer.

Then , simply match the waves of the 2 sources. The last one to attempt anything

like this , I believe, was Wylie Coyote back in the 1960's. But as usual, he wasn't

successful.    " Beep- Beep"

David Downs had a similar thought. Compounding Frequencies? Hmmmm…Back to The Bolts for me. Are you aware of anything that graphically translates the Oscillating Sine wave on a spectral analyzer into Digital Text or Cortext.  What I mean is….Record a voice,  Watch the SINE WAVE on the OSCILLATOR, It make a graph.  Could that graph be in some way translated to compounded notation? Know of any books that might help.  

It is a jumping point.

Kristofer P.D.Q. Emerig said:

Yes the diction is strange, but immensely entertaining.

I believe the theory espoused here is that by compounding enough frequencies of a certain timbre, we can create another desired timbre. This is sort of like one million monkeys typing for a century to try to produce a sensible English sentence by mere chance and volume of typing. The reasoning here is that any particular, arbitrary complexity could be achieved by compounding simpler complexities. The point is, as the time alotted approaches infinity, you can not be certain that the monkeys will type any specified sentence.

My guess is that this would not work, because it is predicated on the assumption that there is but one infinite complexity to which we are converging. Instead, there are multiple infinities diverging. Between any two pitches, there are an infinite array of frequencies which do not overlap with the infinite array of frequencies between two other pitches. Two independent infinities; In fact, an infinity of distinct infinities.

The fact might very well be that the violin does not contain the frequencies you need, in strong enough proportions to undesired ones, so that in the resonant effect of compounding one upon another, you remain just as far, or get further, from the desired timbre as when you started.

I like to play along.

My father and brother were sitting at the breakfast table one morning. My father said to my brother, "If I told you to jump off the back porch would you do it?"  My brother said "yeah" and proceeded to go out the kitchen door to the back porch, which was death defyingly high, A West Virginia house on the side of a mountain. I mean, any one jump off that back porch, broke leg.  But my brother said "yeah" and (He was 6 years old) proceeded to the back porch….My dad in fear he was actually going to jump off and hurt himself. As soon as he stepped onto the porch he swung around facing the kitchen where we all sat, stooped down to put a spring in his little legs, and JUUUMMMPED…..right back into the kitchen.  Dad's point on how my brother was foolish ended that day.

Now, Jump back into the kitchen and please give me a NUTS & BOLTS idea,  no matter how whacky or how pragmatic, that this thing can be done.

Brutishly,  Wesley

michael diemer said:

I suggest we get the National Weather Service working on this. They have been producing high-quality human voice simulations for deacdes now, such as:

"thenationalweatherservicveingraymainehasdetectedalineofseverethunderstormsmovingsouthat45milesperhourthesestormsareassociatedwithariskoftornadoesandseveredownpourspeopleinlowlyingareasareadvisedto takeshelterimmediately"

Wes, you might need to use a 'middleman' here

   I'm talking about a synthesizer (it can imitate in both directions) 

 If a robot can imitate a human voice, then it can learn (be programmed)

to sing . There has to be a way to in a sense 'reverse engineer' this and then

apply it.        Doe's this make any BOLTS or is it just NUTS ?     RS

Is this a Zen koan?

WHY? WHAT DOES WHY HAVE TO DO BUT GET IN MY WAY OF ACCOMPLISHING THIS WONDERFUL IDEA?



Kristofer P.D.Q. Emerig said:

Clearly, you are resolute in taking a practical run at this, so... 

I would research available frequency analysis software, with the hope of finding something which could actually quantify a match between two snippets of soundfile, expressed as a percentage. The violin, playing middle C, for example, might be a 0.000013% match with your speaker pronouncing "ah" for a comparable interval of time.

What would follow is the gradual experimental addition of more and more violins doing randomly tried things. Where an added element increased the percentage match, it would be retained, If not, it would be discarded. This trial and error process would continue until, possibly a century later, the choir of violins enunciated perfectly the phrase "why in the world did we do this?" 
 
Wesley Lawrence Curry II said:

David Downs had a similar thought. Compounding Frequencies? Hmmmm…Back to The Bolts for me. Are you aware of anything that graphically translates the Oscillating Sine wave on a spectral analyzer into Digital Text or Cortext.  What I mean is….Record a voice,  Watch the SINE WAVE on the OSCILLATOR, It make a graph.  Could that graph be in some way translated to compounded notation? Know of any books that might help.  

It is a jumping point.
 

I made my guitar say "I love you" before, but that was a recording that I edited with the EQ after the fact. How to do this live, even with many instruments, would be difficult if not impossible. They would have to be able to sound the different vowel sounds, which might even be harder than the consonants actually. Different vowel sounds require different shapes, if sounding acoustically (like the human voice does). I like the idea, but I'm not sure it could be done with one singular type of instrument without being recorded and equalized.

Whoah, that's some necromancy.

Anyway, this may be relevant to the topic at hand: http://repmus.ircam.fr/orchidee

Reply to Discussion

RSS

Sign up info

Read before you sign up to find out what the requirements are!

Store

© 2020   Created by Gav Brown.   Powered by

Badges  |  Report an Issue  |  Terms of Service