Wish for timbre controller device

arachnaut · Post by **arachnaut** » Sun Sep 17, 2006 11:29 pm

I have a wish for a new device that I can try to describe.

This device would do in the frequency domain what a compressor or
expander does in the time domain. Where they resist or assist dynamics
of intensity change my conceptual device resists or assists dynamics of
timbral change.

For example, if set up in the 'decrease mode' and placed after a swept
resonant filter, it might downsample the harmonics, maintaining an
average pitch of the sound - like an inverse shepard function.

I imagine it as a way of controlling the excitation that might occur in
massive chains of devices.

On the other hand, if set up to assist timbral change, it would generate
some type of harmonic in consonance with the input and thus be a type of
exciter.

Does this sound possible for Live 7?

tylenol · Post by **tylenol** » Mon Sep 18, 2006 8:10 pm

I'm not sure I'm following entirely, but isn't timbre somewhat more complicated than dynamics? In that timbre (the psychoacoustic notion) consists of properties in both the frequency and the time domain? Also, there are aspects of timbre that have representation in the frequency domain but not as harmonics -- i.e. noisiness etc. I think the real problem is that timbre is a psychoacoustic concept, and really has no single acoustic correlate.

arachnaut · Post by **arachnaut** » Mon Sep 18, 2006 8:25 pm

Perhaps an example might make this clearer.

Generally, unless one is very careful, adding more stuff to a
device chain makes the sound muddier.

We try to tame this with filters and compressors or whatever,
but in a dynamic setting where the incoming signal can change
drastically, this can be hard.

While I don't know if my device is feasible, it sees that some
means of simplifying harmonic content that dynamically
adjusts for timbral changes might help.

I don't know the technical, or psychoacoustic terms to be
more specific, but I think of this as a channel sound tamer.

I would like to keep the character of the sound, but remove
unnecessary texture (whatever that means).

If this is possible to make, it's opposite action would take simple
sounds and enrich their harmonic texture.

An example that is somewhat possible:

Suppose you have a very reverberant room recording.
If you had the room's impulse rsponse, you could de-convolve
the signal and perhaps clean up the sound a bit.

blakflag · Post by **blakflag** » Tue Sep 26, 2006 1:28 am

Nice idea, arachnaut. Not sure if anyone has done it before. One obvious implementation would be to do an FFT on the sound to get the frequency bands and then track and mopdify the amplitude changes of each band much as a compressor does for amplitude. Unfortunately think of the processing power needed, you're running the equivalent of a compressor on each band of frequency. If you had 30 frequency bands, then it would be like 30 compressors running plus the overhead of the band splitting. Maybe you could do it.

But it would be a cool research project for some DSP guru anyway.

tylenol · Post by **tylenol** » Sun Oct 01, 2006 1:35 am

Running across these plugins this thread popped into my mind. Especially the spectralcompand plugin -- "performs smooth timbral filtering. Can be used to tame resonant frequencies or for radical changes of tone." I haven't tried them though (can't afford to buy plugins right now

)

arachnaut · Post by **arachnaut** » Mon Oct 02, 2006 2:45 pm

[quote="blakflag"]If you had 30 frequency bands, then it would be like 30 compressors running plus the overhead of the band splitting. [/quote]

It may be far easier. The mud usually only in an octave or two
around the mid-range - maybe from 200-500 Hz.

A few 1/3 octave bands may be enough.