Re: [MMM] Open-source/free production tool for micro and macro-tonal consonance

Carl,
--------------------------------------------------------------------------------------------------------------
>For me personally, I am happy to wait until the docs arrive (I am
>already curious to find out what "1" does...).
1 basically makes a compressed file out of the peaks that's about 1/42nd the original wave file's size (similar compression to AAC-plus/AAC-HE) and, since the encoded file is time-domain, it can be recompressed by LPC-related music encoders like BONK to well under 1/100th the original wave file size and the LPC re-encoding has little to no effect on quality (hence the program's Dot100 name).
Option 0 does decompression. Because of a few phasing errors in resynthesis, though, the decompressed file sounds dull and unsuitable for microtonal or other music production 56-64kb/sec mp3-ish sounding (which is why, unless asked, I won't cover it in too much detail here).
--------------------------------------------------------------------------------------------------------------
Since this whole thing is an invention of sorts (nothing existing so far I've found I can easily compare it how it works) there's a good chance I'm going to use your and other input to develop the final docs for the code to take advantage of more well known terminologies. On demand I may also make an optional Windows version with a complete, simple graphical interface that will in a way document itself, would you (or anyone else) find that useful?

>The "basic" reason it works is it categorizes each instrument part's
>base harmonic and overtones and retunes the overtones to match a
>fairly flexible structure based on tonotopic theory (as opposed to
>"auto-tuning", which forces base harmonics, and not overtones, to
>match an unflexible, often emotionally compromised scale).

>Sounds like some form of adaptive tuning. Are you familiar with
>Bill Sethares' stuff (music/book)?
Far as I've learned it has many similarities but is not as complex. From what I understand Bill's work on tune, timber, and scale (huge oversimplification I know) involves re-tuning each instrument including overtones to a scale to minimize clashing overtones (or scale to instrument) that would bump up disonance between sounds so the new instrument matches the feel of the scale (with disonance only in places that make emotional sense).
My interpretation simply fits existing overtones to a frequency schema that creates a minimum overall dissonance curve between all frequencies.

>My only concern is resynthesis errors -- or are you just pitch-
>shifting bands of audio or something?
It is a form of pitch-shifting only the most valid peak bands. But everything is on a per-window basis, not constant shifts through the whole file (which would sound like a flat and garbled mess)...so nearly all the peaks are preserved. It's really not that complex an algorithm far as DSP perceptual codec-related code goes.

>And where does the binaural part come in?
Oversimplification: it lets the ear sum up parts of music together, rather than duplicating the data in the same channel nearer to another frequency thus making more dissonance.
When electronic musicians of producers pan two different notes to the far left and far right and it sounds clearer they take advantage similar properties.

>Just put your effort into the docs if you prefer -- I can wait to
>have my questions answered there if it's better for you.
It's all good it actually helps to talk things out so I don't end up writing something cryptic in the docs.

---------------------------------
Sponsored Link

$420,000 Mortgage for $1,399/month - Think You Pay Too Much For Your Mortgage? Find Out!

[Non-text portions of this message have been removed]

Hi Michael,

Thanks for the reply. I'm copying this to the main tuning list,
where these more technical discussions are more appropriate (are
you aware of it? /tuning ). If
you like, just reply there.

> Since this whole thing is an invention of sorts (nothing existing
>so far I've found I can easily compare it how it works) there's a good
>chance I'm going to use your and other input to develop the final docs
>for the code to take advantage of more well known terminologies.

Great!

>On demand I may also make an optional Windows version with a complete,
>simple graphical interface that will in a way document itself, would
>you (or anyone else) find that useful?

I would, yes.

>>>The "basic" reason it works is it categorizes each instrument part's
>>>base harmonic and overtones and retunes the overtones to match a
>>>fairly flexible structure based on tonotopic theory (as opposed to
>>>"auto-tuning", which forces base harmonics, and not overtones, to
>>>match an unflexible, often emotionally compromised scale).
>>
>>Sounds like some form of adaptive tuning. Are you familiar with
>>Bill Sethares' stuff (music/book)?
>
> Far as I've learned it has many similarities but is not as
>complex. From what I understand Bill's work on tune, timber, and
>scale (huge oversimplification I know) involves re-tuning each
>instrument including overtones to a scale to minimize clashing
>overtones (or scale to instrument) that would bump up disonance
>between sounds so the new instrument matches the feel of the scale
>(with disonance only in places that make emotional sense).
> My interpretation simply fits existing overtones to a frequency
>schema that creates a minimum overall dissonance curve between all
>frequencies.

Sethares is probably most famous for take a scale as given, then
cooking timbres to minimize tonotopic dissonance for dyads in it
(as I understand it), then composing music with the scale and
timbres. (I don't know if it's any more or less complex than what
you're doing but...) He reads this list from time to time and
maybe he'll try your software and chime in.

>>My only concern is resynthesis errors -- or are you just pitch-
>>shifting bands of audio or something?
> It is a form of pitch-shifting only the most valid peak bands.
>But everything is on a per-window basis, not constant shifts through
>the whole file (which would sound like a flat and garbled mess)...so
>nearly all the peaks are preserved. It's really not that complex an
>algorithm far as DSP perceptual codec-related code goes.

Cool. Bill has done tunes with more local effects (like Three Ears)
than I described above but none using this exact approach that I
know of.

>>And where does the binaural part come in?
> Oversimplification: it lets the ear sum up parts of music
>together, rather than duplicating the data in the same channel nearer
>to another frequency thus making more dissonance.
> When electronic musicians of producers pan two different notes
>to the far left and far right and it sounds clearer they take
>advantage similar properties.

So you move potentially beating peaks to different ears, yes?

Let me just go back to this for a sec:

>The "basic" reason it works is it categorizes each instrument part's
>base harmonic and overtones

Do you mean *fundamental* and overtones, and if so, how do you know
what's what for a polyphonic instrument like a piano?

>and retunes the overtones

...but not the fundamental?

>to match a fairly flexible structure based on tonotopic theory (as
>opposed to "auto-tuning", which forces base harmonics, and not
>overtones, to match an unflexible, often emotionally compromised scale).

I guess I don't know how auto-tuning works -- They have to pitch
shift everything, right (not just the fundamental), but then just
apply a resonance at the formant peaks from the original??

>The reason why recording in different tunings can work together is each
>peak harmonic/overtone is pushed up or down in frequency to align with
>the schema. So, for example, amplitude peaks in the guitar track at
>956hz and the piano track at 1023hz may both be added to produce a peak
>at 1010hz, the nearest point in the schema and their phases rotated
>accordingly to maintain most of the original track's non-noisy
>brightness.

I'll try to try it later today.

-Carl