Kyma Forum
  Kyma Support
  Easiest way to pan polyphonic voices

Post New Topic  Post A Reply
profile | register | preferences | faq | search

next newest topic | next oldest topic
Author Topic:   Easiest way to pan polyphonic voices
cristian_vogel
Member 		posted 05 April 2006 03:15         Edit/Delete Message   Reply w/Quote
Whats the easiest way to pan the different voices of a sound with a MIDIVoice that is making it polyphonic?

IP: Logged

SSC
Administrator 		posted 05 April 2006 08:30         Edit/Delete Message   Reply w/Quote
You could make the panning value dependent on ?VoiceNumber. The Pan or Attenuator used for panning has to be to the left of the MIDIVoice in order for this to work. Then each of the MIDI voices can have a different pan position or offset.

IP: Logged

cristian_vogel
Member 		posted 06 April 2006 06:57         Edit/Delete Message   Reply w/Quote
ok - thanks - how about panning the different tracks of an Oscillator bank|?

[This message has been edited by cristian_vogel (edited 06 April 2006).]

IP: Logged

SSC
Administrator 		posted 06 April 2006 08:54         Edit/Delete Message   Reply w/Quote
The only way is to split it into 2 or more oscillator banks, each covering a different range of the spectrum (you can use the SpectrumModifier to get subsets of tracks).

IP: Logged

cristian_vogel
Member 		posted 07 April 2006 12:19         Edit/Delete Message   Reply w/Quote
cool SSC, thankyou!

IP: Logged

tuscland
Member 		posted 07 April 2006 18:21         Edit/Delete Message   Reply w/Quote

spectrumpanner.kym
 
Hi Cristian,

As the spectrum stream is defined as one sample for the amplitude and one sample for the frequency, you can also select the spectrum tracks with a window that let pass the odd and even tracks for example.
I have made a sound that pan the odd tracks on the left and even tracks on the right, see the attached example.

Cheers,
Camille

IP: Logged

cristian_vogel
Member 		posted 08 April 2006 12:52         Edit/Delete Message   Reply w/Quote
fantastic thankyou - by splitting the tracks like this, it makes the reAnalysis much more interesting to listen to ! Thankyou very much - Its an interesting patch, I have a few questions, they might be silly but ...

Why is one Osicllator bank ( the odds I think) have 56 as a nbrOscillators and the other side has 128? - is that according to some principle or arbitrary DSP control?

Also, can you explain a little about why the control oscillators that you use to filter the result from the SpectralAnalysis , is such a strange frequency value ? a Frequency of 4 samples? Its difficult to grasp Frequency in terms of samples, it would help alot to learn, if you could explain that a little more...

IP: Logged

tuscland
Member 		posted 08 April 2006 13:15         Edit/Delete Message   Reply w/Quote
quote:
Originally posted by cristian_vogel:
fantastic thankyou - by splitting the tracks like this, it makes the reAnalysis much more interesting to listen to ! Thankyou very much - Its an interesting patch, I have a few questions, they might be silly but ...

glad you liked the patch.
you actually gave me the idea. I think there are great things to do with spacial processing and Kyma. I think I'll spend some time on this over the next weeks ...

quote:
Why is one Osicllator bank ( the odds I think) have 56 as a nbrOscillators and the other side has 128? - is that according to some principle or arbitrary DSP control?

good point.
that's a mistake !
I started with a sound that had 56 tracks at the begining and then added a second oscillator to complete the stereo output.
It makes me think that if the spectral input has n tracks, each OscillatorBank much be set with n/2 oscillators.

quote:
Also, can you explain a little about why the control oscillators that you use to filter the result from the SpectralAnalysis , is such a strange frequency value ? a Frequency of 4 samples? Its difficult to grasp Frequency in terms of samples, it would help alot to learn, if you could explain that a little more...

It is actually very simple, as I notes in the previsou post, the stream of a spectral analysis is a sequence of 1 sample for the analyzed amplitude, and 1 sample for the analyzed frequency.
If I want to keep the odd harmonics and reject the even harmonics, then I have to keep 2 samples (odd amp+freq) and reject the following 2 samples (even amp+freq). So I have to multiply 4 samples with a square control signal (that goes from 1 to 0, or the contrary I don't remember).

The first two samples will be multiplied by 1, thus the odd frequencies will pass through.
The next 2 sample will be multiplied by 0, thus the even frequencies are removed.

I can route this modified spectra to an OscillatorBank that plays the original spectra minus even frequencies.

Respectively, if you change the square control signal by changing its phase by 180, you will reject odd frequencies and reject even ones. This different spectra can be routed to another OscillatorBank, thus producing the spacial effect.

Don't hesitate to ask if something is noot clear!

Cheers,
Cam

IP: Logged

pete
Member 		posted 08 April 2006 16:20         Edit/Delete Message   Reply w/Quote
Sorry about being pedantic but all the levels are on the left and all the frequencies are on the right of the spectral stream, so if you change the 4 samp to 2 samp you then get the odds on one side and the evens on the other. As it is you are getting 1,2,5,6,9,10 etc on one side and 3,4,7,8,11,12 etc on the other. but it's a dam good sound either way.
Thanks

[This message has been edited by pete (edited 08 April 2006).]

IP: Logged

tuscland
Member 		posted 08 April 2006 18:04         Edit/Delete Message   Reply w/Quote
Second mistake !
Thanks for the correction, and yes it sounds really more logical and simple this way.

So that means I was having frequencies 1/2, 5/6, 9/10 ... on one side and 3/4, 7/8, 11/12 ... on the other side. Funny :-)

Thanks Pete,
Cam

IP: Logged

tuscland
Member 		posted 08 April 2006 18:17         Edit/Delete Message   Reply w/Quote
Hi Pete,

An update of my previous post :
If I read the documentation of the SpectrumInRAM prototype, it is said that a spectrum is as a sequence of (amplitude,frequency) pairs on every sample tick for nbrPartials samples. It doesn't state that the amplitude and frequency domains are outputted on both channels.

I think you were actually talking about the FFT output.

Am I right of the definition of SpectrumInRAM has changed ?

Cheers,
Cam

IP: Logged

SSC
Administrator 		posted 08 April 2006 18:47         Edit/Delete Message   Reply w/Quote
In the Spectrum Sources, the amplitude of each pair is on the left channel; the pitch of each pair is on the right channel. On each sample tick, one partial (one amp/pch pair) is output. The length of a "frame" is the number of partials in your analysis. For example, if your analysis has 128 partials, a frame lasts 128 samples.

IP: Logged

tuscland
Member 		posted 09 April 2006 03:33         Edit/Delete Message   Reply w/Quote
Thanks for the clarification !

[This message has been edited by tuscland (edited 09 April 2006).]

IP: Logged

cristian_vogel
Member 		posted 12 April 2006 04:51         Edit/Delete Message   Reply w/Quote
MIDIVoice alternate voice panner

this seems to work and is quite neat

in the Left box
?VoiceNumber mod: 2

in the RIght
(1 + ?VoiceNumber) mod: 2


IP: Logged

All times are CT (US)

next newest topic | next oldest topic

Administrative Options: Close Topic | Archive/Move | Delete Topic
Post New Topic  Post A Reply

Contact Us | Symbolic Sound Home

This forum is provided solely for the support and edification of the customers of Symbolic Sound Corporation.


Ultimate Bulletin Board 5.45c