Hello listers~
There's been a void of conversation on the list, so I thought I'd pitch my latest idea for a Pd tool: a granular cross-fader. I'm in the pre-planning phase right now, and would love to get some ideas/suggestions for how to implement this.
Basically what I'd like to have is similar to a DJ cross fader, but using granular methods to stochastically mix two (or more) signals using various common granular ideas (pitch shifting, time stretching, grain size, randomness parameters, overlap, etc.). This would make transitions much more exciting from a DJ end as well as when mixing two or more signals on a DAW.
What are your immediate thoughts/reactions to this?
~Kyle
Kyle Klipowicz a écrit :
Hello listers~
There's been a void of conversation on the list, so I thought I'd pitch my latest idea for a Pd tool: a granular cross-fader.
Why granular? Why not a convoluting crossfader rather?
___________________________________________________________________________ Yahoo! Mail réinvente le mail ! Découvrez le nouveau Yahoo! Mail et son interface révolutionnaire. http://fr.mail.yahoo.com
Oh shoot... can you explain that more? I understand convolution in reverb but how might that work in a crossfader?
km
Why granular? Why not a convoluting crossfader rather?
Yes, please do. That sounds fascinating!
~Kyle
On 1/12/07, Kevin McCoy km.takewithyou@gmail.com wrote:
Oh shoot... can you explain that more? I understand convolution in reverb but how might that work in a crossfader?
km
Why granular? Why not a convoluting crossfader rather?
--
++++ http://pocketkm.blogspot.com
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Kevin McCoy a écrit :
Oh shoot... can you explain that more? I understand convolution in reverb but how might that work in a crossfader?
km
Why granular? Why not a convoluting crossfader rather?
The idea is about filtering one track with the spectral enveloppe of the other one, and vice-versa, The crossfade would pass progressively from one track to another, but instead of having mixed tracks at the center, there would have the convolution between the tracks: There is a bad side, because this kind of effect can only affect the filtered sound in real time.
There is an approach of this in I06.Timbre.stamp.pd
PC.
___________________________________________________________________________ Yahoo! Mail réinvente le mail ! Découvrez le nouveau Yahoo! Mail et son interface révolutionnaire. http://fr.mail.yahoo.com
I just spent the afternoon trying to achieve this effect, with little success. Here is what I've been thinking, hopefully someone knows where I am going wrong, or has some information they can share on the matter.
There needs to be a fade-in of the spectral envelope applied to signal 1, ranging from all the fft bins equal to 1 when signal 2 is not to be heard to the fft magnitude of signal 2 (which is what the timbre.stamp.pd patch does) when the two are to be cross-synthesized. If the fade is supposed to go all the way until only signal 2 is heard, then the spectral envelope fade needs to be switched to apply to signal 1.
Is there an easier way to do it? My attempt (attached) at doing only the first part (fading in the spectral envelope of signal 2 applied to signal 1) barely sounds more than a regular crossfade until the regular timbre stamping happens when the fade-in goes to 1. What I'd like to have is something that took more of it's 'timbre' from one signal than the other, and be able to fade this amount continuously.
regards rich
On 1/12/07, Patco megalegoland@yahoo.fr wrote:
Kevin McCoy a écrit :
Oh shoot... can you explain that more? I understand convolution in reverb but how might that work in a crossfader?
km
Why granular? Why not a convoluting crossfader rather?
The idea is about filtering one track with the spectral enveloppe of the other one, and vice-versa, The crossfade would pass progressively from one track to another, but instead of having mixed tracks at the center, there would have the convolution between the tracks: There is a bad side, because this kind of effect can only affect the filtered sound in real time.
There is an approach of this in I06.Timbre.stamp.pd
PC.
Yahoo! Mail réinvente le mail ! Découvrez le nouveau Yahoo! Mail et son interface révolutionnaire. http://fr.mail.yahoo.com
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Sounds good Kyle.
If you listen to old skool house DJs they often use the tone control to start with only the hi-hats/top of the mix (because that is where you get the best time acuity) then as the mix fades to the second part they drop in the bass.
For the mostpart only a few parameters will work for the grainsize and randomness without it sounding watery and blurred, an optimal window size and overlap.
The starting assumption is that you have two audio files and two phasors/line segments that give you a pair of timelines that are locked in beat sync.
The most obvious "granular mix" effect is to mix N grains from one side with 1-N grains from the other per unit time.
But the weight/density of the sound is a product of the number of grains per unit time and the grain size, so another mix would be to increase the size of one grain stream and reduce the other.
You could make another interesting effect by "dissolving" one stream into chaos by adding random time offsets to blur it out, and then quickly mixing the other one in with the same amount of blurring, then bring the new stream back into focus.
As far as tone/spectrum goes, smaller grains will generally be higher (because they must contain shorter periods) and longer grains will have more low frequencies.
Either way, you'll get modulation artifacts for grains whos period is less than 1/F if F is the lowest frequency in the mix, but I assume that is okay, just part of the effect.
On Thu, 11 Jan 2007 13:53:01 -0600 "Kyle Klipowicz" kyleklip@gmail.com wrote:
Hello listers~
There's been a void of conversation on the list, so I thought I'd pitch my latest idea for a Pd tool: a granular cross-fader. I'm in the pre-planning phase right now, and would love to get some ideas/suggestions for how to implement this.
Basically what I'd like to have is similar to a DJ cross fader, but using granular methods to stochastically mix two (or more) signals using various common granular ideas (pitch shifting, time stretching, grain size, randomness parameters, overlap, etc.). This would make transitions much more exciting from a DJ end as well as when mixing two or more signals on a DAW.
What are your immediate thoughts/reactions to this?
~Kyle
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Good feedback!
I'm not so concerned about beat matching with this thing as I am transitioning from multiple sections in a mix using Ableton. I'm more just wanting to have say a vocal/acoustic guitar verse slowly blend together and then (granularly) fade into a drumset/keyboard chorus or something like that. My aesthetic right now is very much one of "pop structure becomes abandoned into a chaotic transition into another pop structure." Rinse and repeat as you'd like.
Do you think this could be done in real time in an efficient fashion, using a kind of doubled up granular delay network?
~Kyle
On 1/12/07, padawan12 padawan12@obiwannabe.co.uk wrote:
Sounds good Kyle.
If you listen to old skool house DJs they often use the tone control to start with only the hi-hats/top of the mix (because that is where you get the best time acuity) then as the mix fades to the second part they drop in the bass.
For the mostpart only a few parameters will work for the grainsize and randomness without it sounding watery and blurred, an optimal window size and overlap.
The starting assumption is that you have two audio files and two phasors/line segments that give you a pair of timelines that are locked in beat sync.
The most obvious "granular mix" effect is to mix N grains from one side with 1-N grains from the other per unit time.
But the weight/density of the sound is a product of the number of grains per unit time and the grain size, so another mix would be to increase the size of one grain stream and reduce the other.
You could make another interesting effect by "dissolving" one stream into chaos by adding random time offsets to blur it out, and then quickly mixing the other one in with the same amount of blurring, then bring the new stream back into focus.
As far as tone/spectrum goes, smaller grains will generally be higher (because they must contain shorter periods) and longer grains will have more low frequencies.
Either way, you'll get modulation artifacts for grains whos period is less than 1/F if F is the lowest frequency in the mix, but I assume that is okay, just part of the effect.
On Thu, 11 Jan 2007 13:53:01 -0600 "Kyle Klipowicz" kyleklip@gmail.com wrote:
Hello listers~
There's been a void of conversation on the list, so I thought I'd pitch my latest idea for a Pd tool: a granular cross-fader. I'm in the pre-planning phase right now, and would love to get some ideas/suggestions for how to implement this.
Basically what I'd like to have is similar to a DJ cross fader, but using granular methods to stochastically mix two (or more) signals using various common granular ideas (pitch shifting, time stretching, grain size, randomness parameters, overlap, etc.). This would make transitions much more exciting from a DJ end as well as when mixing two or more signals on a DAW.
What are your immediate thoughts/reactions to this?
~Kyle
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
http://www.dafx.ca/slides/poster_067.pdf
this is from IRCAM and it is beat-synchronised, so just to give another idea....
P
Kyle Klipowicz wrote:
Hello listers~
There's been a void of conversation on the list, so I thought I'd pitch my latest idea for a Pd tool: a granular cross-fader. I'm in the pre-planning phase right now, and would love to get some ideas/suggestions for how to implement this.
Basically what I'd like to have is similar to a DJ cross fader, but using granular methods to stochastically mix two (or more) signals using various common granular ideas (pitch shifting, time stretching, grain size, randomness parameters, overlap, etc.). This would make transitions much more exciting from a DJ end as well as when mixing two or more signals on a DAW.
What are your immediate thoughts/reactions to this?
~Kyle
-
Kevin McCoy -
Kyle Klipowicz -
padawan12 -
Patco -
Peter Plessas -
Rich E