----- Original Message -----

From: Nathan Holmes

To: CSOUND@LISTSERV.HEANET.IE

Sent: Wednesday, May 10, 2017 8:28 AM

Subject: Re: [Csnd] can we constitute sound into taxonomic fields?

Perhaps this will be more entertaining than illuminating, but you all might be interested in this Google AI Project, "The Infinite Drum Machine", which used machine learning ("t-distributed stochastic neighbor embedding", apparently) to group audio files together based on similarity.

https://aiexperiments.withgoogle.com/drum-machine

You can quickly browse through the sample set by clicking "Launch Experiment" and then dragging the circles around on the map. Some of it is grouped together pretty nicely: there'a large blue cluster with clinking, glass-like and metallic sounds; and then as you move to the right toward the purple/magenta, the sounds get more clunky and wooden; and the large teal/green sections at the top seems to be more "scrape-y" or have more breath/air present, with the teal stuff on the upper-middle left getting closest to white noise. Or so it seems to me.

(Note, the sounds seem to have a bunch of tags on them, but the project description says the program categorized based on audio alone.)

So, naturally there are occasional samples that seem oddly out of place, but otherwise I find a lot of pleasant coherency in how these sounds are grouped together. Maybe a more refined version of this process could help suggest some overarching categories for timbre?

But if not, it's at least fun to play around with. :)

On Tue, May 9, 2017 at 3:39 PM, Andrea Strappa <a_strappa@tin.it> wrote:

Good, to find out whether the scheme is good one has to work with it.

But first we need to ask: why do you want categorizing the timbre?

To baste a your composition form? Or to teach music to children? Or to analyze a famous composer work? Or to do a statistic about something? Or something?

Then, you might notice the opposition of terms: the “difference between units on the level of expression”.

You can notice that some difference is pertinent in a context, not in other context.

For an Inuit there are a lot of types of snow, with a lot of names, because they live in the snow. I have a few words to describe the state of the snow, because I see the snow a few day every year.

Someone said that this musical era is the era of the timbre.

It’s understandeble the need to reflect about a categorization.

Andrea S.

 

----- Original Message -----

From: Aaron Krister Johnson

To: CSOUND@LISTSERV.HEANET.IE

Sent: Tuesday, May 09, 2017 5:49 PM

Subject: Re: [Csnd] can we constitute sound into taxonomic fields?

 

These are interesting ideas.

 

My initial thoughts for categorizing timbre would be an N-dimensional system (however abstract, but ideally, mapped to quantifiable parameters) that would definitely include:

 

1) harmonicity/inharmonicity of excitation source

1a) fundamental pitch range

2) spectral profile, perhaps represented as a generalized filter contour.

3) attack, decay, sustain, release, etc. (envelope behavior over time)

4) Spacial or "effects" profile

 

Perhaps unconsciously, I'm thinking that, primitive in that ways that it is, the traditional subtractive analog hardwired synth does a good job of modeling a small subset of the possibilities, which could be extended (by analogy or expansion of parametric possibility) to include many natural and 'newer' types of sound sources:

 

    generator(osc/noise,phsyical model, actual physical) -->

    optional spectral shaping via filters -->

    amplitude characteristic shaping (envelope) -->

    wrapping the sound in a "space" or "effects space".

In a naive sense, any synth's timbre space is N-dimensional, and 'N' is equal to the number of knobs + switches in its possible parameters!

I tend to think all timbres I can think of can be categorized as such....even when you have FFT and spectral synthesis and resynthesis, you are still working with the idea of an 'excited' sound source, and if the sound is a natural sound (struck/blown/friction), it is still being shaped (filtered/reverberated) by the physics around it.

 

Acoustic instruments are categorized by Hornbostel/Sachs as 'ideophone - membranophone - chordophone - aerophone - electrophone'; and are strike-/pluck-/blow-/friction- activated. Those categories, in my mind, map to envelopes, both spectral and amplitundinal.

 

-AKJ

 

Csound mailing list Csound@listserv.heanet.ie https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to https://github.com/csound/csound/issues Discussions of bugs and features can be posted here

Csound mailing list Csound@listserv.heanet.ie https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to https://github.com/csound/csound/issues Discussions of bugs and features can be posted here

Questa e-mail è stata controllata per individuare virus con Avast antivirus. www.avast.com