I finally managed to do natural sounding real-time granular time stretching / pitch shifting in Max, with pretty much the same quality as applications like Ableton Live and Reaktor. Unlike FFT-based (phase vocoder) approaches, granular time stretching has no inherent latency.

http://www.arttech.nl/max/time%20stretcher%20080423.zip (note, the patch is in Max 5 format)

This patch uses the great pitch~ object by CNMAT, that you can download here:
http://cnmat.berkeley.edu/downloads (scroll down to the alphabetical list of Individual Downloads and look for pitch~).

Enjoy!

Mattijs

Dan

At 6:06 PM +0200 4/23/08, Mattijs Kneppers wrote:
>Hi all,
>
>I finally managed to do natural sounding real-time granular time
>stretching / pitch shifting in Max, with pretty much the same
>quality as applications like Ableton Live and Reaktor. Unlike
>FFT-based (phase vocoder) approaches, granular time stretching has
>no inherent latency.
>
>http://www.arttech.nl/max/time%20stretcher%20080423.zip (note, the
>patch is in Max 5 format)
>
>This patch uses the great pitch~ object by CNMAT, that you can download here:
>http://cnmat.berkeley.edu/downloads (scroll down to the alphabetical
>list of Individual Downloads and look for pitch~).

–
Dan Nigrin
Defective Records
202 Hack / PC-1600 User / VSTi Host / OMS Convert / Jack OS X / Major
Malfunction
http://www.defectiverecords.com

http://www.jackosx.com

I’d love to hear it, but it doesn’t produce any sound here, I loaded one
of the sample sound files, and don’t know what it supposed to do,
hitting trigger doesn’t create any result, and the numberbox aside has
no lable (no idea what its for…) When loading the beat file, the
waveform~ shows only very small low bits, it seems it doesn’t import the
wav correctly, the Tones example at least looks like a waveform~…
It would be helpful if all number boxes would be limited to useful
values, as I have no idea what a useful value could be…

But anyway, thanks for sharing, I might be able to debug it myself, but
I am preparing a move, so I’d only do it in a few weeks…

Stefan

–
Stefan Tiedje————x——-
–_____———–|————–
–(_|_ —-|—–|—–()——-
– _|_)—-|—–()————–
———-()——–www.ccmix.com

really nice patch. could i ask where the herat sample is from?

cheers,

oli

very exciting. thanks!

Stefan, I can’t reproduce that here. The patch should start playing directly after you read a sample, no need for triggering. Assuming you enabled dsp, of course. Do the samples play correctly in iTunes, for example? Which os/max are you on? I tested on Mac OS 10.4.11, Max 5 34567.

I agree that the number boxes should be limited, thanks for the suggestion. The number box left to the trigger is the position in milliseconds (which you can set by clicking in the waveform) that you can trigger by pressing the bang or the space bar. I added the trigger somewhat at the end to explicitly show the benefits of a granular based approach. Triggering like that, without any latency, is not possible with an fft based system.

Oli, I got the herat sample from a sound sharing website (http://freesound.iua.upf.edu/). It is made by ERH (http://www.myspace.com/sinuspiralmusic).

Best,
Mattijs

And Nick :)

interesting patch but it doesn’t produce any sound for me…

Please help me understand why.

All the best

–
Alessandro Fogar

http://www.fogar.it

2008/4/23, Mattijs Kneppers :
>
> Hi all,
>
> I finally managed to do natural sounding real-time granular time stretching / pitch shifting in Max, with pretty much the same quality as applications like Ableton Live and Reaktor. Unlike FFT-based (phase vocoder) approaches, granular time stretching has no inherent latency.
>
> http://www.arttech.nl/max/time%20stretcher%20080423.zip (note, the patch is in Max 5 format)
>
> This patch uses the great pitch~ object by CNMAT, that you can download here:
> http://cnmat.berkeley.edu/downloads (scroll down to the alphabetical list of Individual Downloads and look for pitch~).
>
> Enjoy!
>
> Mattijs
>

don’t know if this can help but I’m on a G4/ OS 10.4.10

All the best

–
Alessandro Fogar

http://www.fogar.it

2008/4/23, Mattijs Kneppers :
>
> Thanks Dan and Oli!
>
> Stefan, I can’t reproduce that here. The patch should start playing directly after you read a sample, no need for triggering. Assuming you enabled dsp, of course. Do the samples play correctly in iTunes, for example? Which os/max are you on? I tested on Mac OS 10.4.11, Max 5 34567.
>
> I agree that the number boxes should be limited, thanks for the suggestion. The number box left to the trigger is the position in milliseconds (which you can set by clicking in the waveform) that you can trigger by pressing the bang or the space bar. I added the trigger somewhat at the end to explicitly show the benefits of a granular based approach. Triggering like that, without any latency, is not possible with an fft based system.
>
> Oli, I got the herat sample from a sound sharing website (http://freesound.iua.upf.edu/). It is made by ERH (http://www.myspace.com/sinuspiralmusic).
>
> Best,
>
> Mattijs
>
>

I would really like to hear your patch but have
not had the chance to install Max5 yet. I figure
I was going to wait until they fix any missed
quirks. Can you post a Max4 version as well?

I heard some rumors that fiddle~ doesn’t work well with Max 5. But I didn’t try, I’ll look into that. Note that the Beats mode doesn’t use pitch detection, it should work without the pitch~ object.

>
> I would really like to hear your patch but have
> not had the chance to install Max5 yet. I figure
> I was going to wait until they fix any missed
> quirks. Can you post a Max4 version as well?
>

Is that because you want your 30 days to start later or because you’re afraid you can’t use Max 4 anymore after installing Max 5? In case of the latter: Max 5 runs alongside Max 4 perfectly well.

There is no automatic way to convert Max 5 patches to Max 4, but I got a similar request in the context of a project I’m working on so chances are that I’ll do this somewhere in the next month. Or maybe you can find someone who has max 5 installed to do it for you.

Mattijs

Ok, I heard 10.4.11 is necessary for some Max 5 features, especially documentation stuff though.

Best,
Mattijs

don’t think this is related but I’ll upgdare as soon as possible.

Yesterday I investigate better about this problem with the Max5 audio
probes and noticed some problem, will look better later tonite.

All the best

–
Alessandro Fogar

http://www.fogar.it

2008/4/24, Mattijs Kneppers :
>
> Quote: sfogar wrote on Thu, 24 April 2008 00:18
> —————————————————-
> > Hi,
> >
> > don’t know if this can help but I’m on a G4/ OS 10.4.10
> >
> > All the best
> >
> > –
> > Alessandro Fogar
> >
> > http://www.fogar.it
> >
>
> Ok, I heard 10.4.11 is necessary for some Max 5 features, especially documentation stuff though.
>
> Best,
> Mattijs
>

I tried it also on Windows Xp and it works (in beat mode).

What I think it does not work for me on OS X is the soundfiles loading.

Probably this has to do with Quicktime.

All the best

–
Alessandro Fogar

http://www.fogar.it

2008/4/24, Alessandro Fogar :
> Hi,
>
> don’t think this is related but I’ll upgdare as soon as possible.
>
> Yesterday I investigate better about this problem with the Max5 audio
> probes and noticed some problem, will look better later tonite.
>
> All the best
>
> –
> Alessandro Fogar
>
> http://www.fogar.it
>
>
> 2008/4/24, Mattijs Kneppers :
> >
> > Quote: sfogar wrote on Thu, 24 April 2008 00:18
> > —————————————————-
> > > Hi,
> > >
> > > don’t know if this can help but I’m on a G4/ OS 10.4.10
> > >
> > > All the best
> > >
> > > –
> > > Alessandro Fogar
> > >
> > > http://www.fogar.it
> > >
> >
> > Ok, I heard 10.4.11 is necessary for some Max 5 features, especially documentation stuff though.
> >
> > Best,
> > Mattijs
> >
>

Is it based on granuLaRrrr?

Tristan Jehan ported pitch~ to Windows I believe:

http://web.media.mit.edu/~tristan/

Dan
–
Dan Nigrin
Defective Records
202 Hack / PC-1600 User / VSTi Host / OMS Convert / Jack OS X / Major
Malfunction
http://www.defectiverecords.com

http://www.jackosx.com

For those for whom it did not work, you did download and install the CNMAT object [pitch~], yes?

Yes, indeed. I’m getting this extremely distorted, rhythmic ‘choppy’ sound with the sample being barely discernible. I have tried multiple sample rates, overdrive on/off, vector sizes, etc.

if you look at the waveform with an external editor, it’s not the same
which the waveform object in Max5 displays. So I think that the sample
does not load correctly.

All the best

–
Alessandro Fogar

http://www.fogar.it

2008/4/24, matt :
>
> Quote: smill wrote on Thu, 24 April 2008 07:51
> —————————————————-
> > Works just fine for me – MacBook Pro 2.16 GHz, MaxMSP 5, OSX 10.4.11. Nice work, and very much appreciated!
> >
> > For those for whom it did not work, you did download and install the CNMAT object [pitch~], yes?
> —————————————————-
>
> Yes, indeed. I’m getting this extremely distorted, rhythmic ‘choppy’ sound with the sample being barely discernible. I have tried multiple sample rates, overdrive on/off, vector sizes, etc.
> –
> http://www.virb.com/cebec
>

I think this is a Max 5 bug that concerns the import function of buffer~. I tried on two more computers. There was no problem on a MacPro Quad, but there was on a Dual G5, both on 10.4.11. I don’t know if the architecture has anything to do with it though.

Anyway, this is what happens on the ppc:

http://www.arttech.nl/max5/import-probs.mov

As you can see, the file loads fine with the read command but something weird happens when using the import command.

I attached the small test patch in a zip.

Cycling, I’m curious whether you can reproduce this? Note that it seems to work on some computers but not on others..

Best,
Mattijs