Wanted: Developer for time stretch plug

[shamann]

Well a granular effect can be used to time-stretch, but it isn't a pristine kind of time stretching and has limited applications. You could build something in Csound, Supercollider or PD that would work.

What process does Ableton use for its time-stretch? Some kind of PSOLA thing, or are they doing some voodoo that's far ahead of the game?

[pHz]

dont know precisely but there seems to be all kinds of different gubbins under lives timestretching hood ... its just described as 'proprietry' technology so dont know whether they use their own algorhythms or builds on stuff already available like PSOLA (id GUESS the former though - they dont seem like the types to borrow other stuff if they thought they could do it properly themselves) ...

... whatever they do it works BLOODY well though ...

slainte rob

[Galleoneer]

The timestreching alg in ableton (i am talking abou v3) is not so good like the one in the fl studio. what makes it rather complicated to programm is abletons concept of warp markers (!), envelopes and the other (plenty other) parameters. besides programming the dsp functionality. Making the gui with the markers & envelopes etc. would take a developer more the 10 hours (or maybe I am a slow developer ).

[pHz]

lives timestretching has moved on in many ways since live3 ... the new 'complex' mode is pretty advanced ... having said that doesnt fl-studio now use the elastique algorhythm which is accepted as a DAMN good system anyway ???

slainte rob

[;-)]

But, I think most developers here are quiet able to do your job.

...

Something between 3000€-5000€ would do for a start.

[Galleoneer]

But, I think most developers here are quiet able to do your job.

...

Something between 3000€-5000€ would do for a start.

Which part is so funny Jens?
a. most developers here are quiet able to do your job.
or
b. Something between 3000€-5000€ would do for a start.

if its b. then I must say that it will take approx. 2 months to make it plus testing. So for a dipl. ing. in germany this is a not so lucrative estimation. But it is big cash in china. So depending were you live is the correct formulation.

if its a and you think that the devs here a all morons then this thread will become very interesting.

[;-)]

@Exit Zero:

Of course it is possible. But it is also somehow more complicated than you may think.

Such an effect must have an internal beat detector (and possibly pitch detector too) and a whole brunch of different algorithms, encapsulated by a heavy selection logic. Otherwise, the user is responsible for doing it all right (or even wrong) with allot of parameters and special options.

A drumloop for instance has to be streched completely differnt than a single human voice or even polyphonic material, if you want any qualitative (useful) result.

Drumloops merely work right with beat slicing.
Monophonic voices have to be processed with highly specialized and complex algorithms like PSOLA to keep the formants and transients intact.
Polyphonic material can only be stretched satisfactory with FFT/iFFT (resynthesis).

Take a look at Melodyne. Then you may get a raw idea, how complex that entire thingy is...

Finally you probably would end up with another new plugin (about 15 MB) if doing that all yourselfes ...


Intakt is specialized for beat slicing. Althought it has other algorithms (from Kontakt) inbuilt too (and a whole brunch of fancy options to refine the results). It is surely the "state of the art" for the things you want to do ...

Rephrase may be ideal, if you want to work excessively with monophonic voices (like single accoustic instruments and human voices). Intakt wouldn't be able to do that.

Poyphonic material seems to be the simplest case and can be processed easiely with a phase vocoder. Althought, there may be some heavy processor load neccessary to get best quality.


The soundtouch library you mentioned gives you a very average quality for some reasons, because it is obviously too simple in its conception (using a granular technique). Therefore the final result is (mostly) quite bad - especially with drum loops or when used as a DJ mixing tool. But it is "fast" at least.


Believe me: It that case it is always better, to trust a development team, that has years of experiences in the entire area, rather than to do it from scratch ...


Alternatives?
Take a look at the DIRAC time stretch/pitch shift library. It is available with some (bad) limits for free. But probably possible to do a "simple" plugin that way if you can bear those limitations ...

But you can also buy a license of that library to remove the limitations . But don't be surprised by the prices it actually has ...

.

[;-)]

But, I think most developers here are quiet able to do your job.

...

Something between 3000€-5000€ would do for a start.

Which part is so funny Jens?
a. most developers here are quiet able to do your job.
or
b. Something between 3000€-5000€ would do for a start.

if its b. then I must say that it will take approx. 2 months to make it plus testing. So for a dipl. ing. in germany this is a not so lucrative estimation. But it is big cash in china. So depending were you live is the correct formulation.

if its a and you think that the devs here a all morons then this thread will become very interesting.

Who is Jens, by the way?

I could program a ready to use BETA version of such a plug inside 2 days (as with the Detunator), including a complete GUI.

It would have approximately the quality which is achieveble with the SoundTouch library (but I would not use that for some reasons).

But I fear, I (personally) would not be satisfied by even that quality. Because: The principle SoundTouch (and my quick timestretch hack then) uses is basicly a granular technique. It is quite fast and could be performed in realtime with low latency, thus being completely time domain ...

Whatsoever, the results are not that satisfactory when for instance applied to drum loops or solistic voices for the reasons I described in my last post. And the user has to input correct initial values to specify the base playback speed prior triggering it ...

But it would be able to react continuously to tempo changes and modify pitch / timestretch in realtime ...


I have the clear opinion, that a really professional solution (I always have those claims to quality) would require allot more of investigation and time. And money, of course .

But I also would not waste any time to "re-invent the wheel" with doing that, because those tools are already available. Highly specialized, with pristine quality and quite numerous, by the way.

.

[Scotrick]

Isn't Melodyne Uno, sort of like this? I mention is only as an example of how hard and expensive it must be to take a single MONO audio part and alter it

[jens]

CodeAudioBeatburner might do the trick for you as well, though it is intended more as an audio mangler. Well you only have to mangle so much as you like ...

no time-stretching..

[Galleoneer]

Who is Jens, by the way?

according to some guys here at KvR intelligence (I can't tell you the names of my informants without killing you afterwards ) is jens the name of the creature who programmed detunator. You claim to have programmed it -> you must be jens (simple?).

I could program a ready to use BETA version of such a plug inside 2 days (as with the Detunator), including a complete GUI.

I doubt it. Look, he wants the same thing as in ableton live. and the live timestretcher sampleplayer works with warp markers and envelopes that have a quite complex gui. Only the gui would take you more than 2 days. Maybe you could do the core dsp stuff in two days, if you have the needed components, but not the entire thing with gui

[shamann]

I think Jens is a second developer. Looks like there's more than one developer over at dizainer.net anyway.

What do the warp markers do exactly?

[;-)]

I could program a ready to use BETA version of such a plug inside 2 days (as with the Detunator), including a complete GUI.

I doubt it. Look, he wants the same thing as in ableton live. and the live timestretcher sampleplayer works with warp markers and envelopes that have a quite complex gui. Only the gui would take you more than 2 days. Maybe you could do the core dsp stuff in two days, if you have the needed components, but not the entire thing with gui

Wrong.
He actually wrote:

"I want something with very simple functionality, low CPU hit, but rock solid stability for live performance. It could handle just 1 loop for all I care and I could run multiple instances."

That's all. Nothing of "complex GUI with worp markers and envolver leppos" or similar nonsense. Just simply loading a loop and realtime stretching it.

But of course, you can doubt what you want! Doubting is quite easy (rather than doing something).

.

[pHz]

Wrong.
He actually wrote:
"I want something with very simple functionality, low CPU hit, but rock solid stability for live performance. It could handle just 1 loop for all I care and I could run multiple instances."
That's all. Nothing of "complex GUI with worp markers and envolver leppos" or similar nonsense.

he ALSO wrote ...

handle looping playback of a given wav file in the same fashion that Ableton live does?

... and (admittedly less pertintently) ...

I prefer to have something using time stretching, like Acid or Ableton Live.

But of course, you can doubt what you want! Doubting is quite easy. .

i doubt it ...

slainte rob

[;-)]

Wrong.
He actually wrote:
"I want something with very simple functionality, low CPU hit, but rock solid stability for live performance. It could handle just 1 loop for all I care and I could run multiple instances."
That's all. Nothing of "complex GUI with worp markers and envolver leppos" or similar nonsense.

he ALSO wrote ...

handle looping playback of a given wav file in the same fashion that Ableton live does?

... and (admittedly less pertintently) ...

I prefer to have something using time stretching, like Acid or Ableton Live.

But of course, you can doubt what you want! Doubting is quite easy. .

i doubt it ...

slainte rob

I know, you are specialist in "mangeling" information, rising it out of any context and trying to make it fit your needs...

Whatsoever here is the whole post again. Try to read it in context:

Why is there no VSTI plug currently that can handle looping playback of a given wav file in the same fashion that Ableton live does?

There are plugs (like Xtal) which can pitch and time shift for a given *static* tempo.

However, I need a plug that can do the same thing but with *dynamic* tempo changes. The playback should stay in sync while tempo is being adjusted (like it does in Ableton Live.)

There are plugs which use REX-like beat chopping (Microdicer, E-lab Obsession, etc.) but I prefer to have something using time stretching, like Acid or Ableton Live.

NI Intakt can do this, but the plug is 15mb!

I want something with very simple functionality, low CPU hit, but rock solid stability for live performance. It could handle just 1 loop for all I care and I could run multiple instances.

Would there be a kind developer here who would be interested in this project? I would certainly consider paying a developer if they can meet my needs.

Xtal uses the soundtouch library for time stretching, so it may be of some interest: http://sky.prohosting.com/oparviai/soundtouch/

Also, some enlightening discussions about Xtal, Energy XT and the need for a time strecthing plugin can be read on this thread as well:
http://www.kvraudio.com/forum/viewtopic ... =0&postord er=asc&highlight=xtal&start=0

But probably you are not able to extract the kernel of the information/question never the less ...

.