On April 14, 2013 01:04:19 PM Robert Jonsson wrote: > Hi Florian, > > nice to see you :) > > 2013/4/12 Florian Jung <[email protected]>: > >>> I have a suggestion. > >>> I think it would be simple to replace the off-line resampler code with > >>> > >>> RubberBand's off-line functions or something else instead of just SRC. > >>> > >>> What was that nice one Flo mentioned? > >> > >> Ah, indeed. stretching would improve it a great deal. I took a look at > >> an example I found in the rubberband sources but it looked awfully > >> convoluted. Is there an offline version of the api? > >> Need a clearer head before trying to adapt that code.. though maybe > >> there are other examples that are easier to adapt. > > > > i *might* be able to help you with these stretchers, i touched all of > > them some months ago. i'm not fully sure, but i think rubberband offers > > an offline mode. setKeyFrameMap is the point to poke. > > Okay, if you know an example do send it. > Realtime resampling would surely be nice but for the moment it's > overkill in Simple Drums so I'd prefer starting small :) > > >>> By carefully setting their pitch and length controls I think these > >>> shifters > >>> > >>> can also replace SRC to do the resampling job on non-MusE sample rate > >>> files.>> > >> Yes, as long as the resampling is 'good enough' it should not matter > >> if we are stretching or changing sample rate. > > > > i think, we should consider doing resampling and time-stretching in the > > same step anyway. I'd like to have pitch-shifting as well in MusE, and > > pitch-shifting is nothing else than time-stretching and then resample. > > Yeah, that would be preferrable. > Tim, you mentioned something about time-stretching not being good > enough for resampling. Does that mean there a noticable quality > difference between secret rabbit code and rubberband?
Well I read some snippet that said it is generally better to use SRC for resampling than using a stretcher, but I have no personal experience to verify this. On the other hand that was in reference to RubberBand, maybe another stretched would be better at it? Here is a line from QTractor's feature page: "Audio clip time-stretching (WSOLA-like or via librubberband), pitch-shifting (via librubberband) and seamless sample-rate conversion (via libsamplerate)." But then, here is a line from RubberBand's page: "... it requires separate library support for the FFT and resampling implementations: for the Free Software edition, this means FFTW and libsamplerate (the proprietary edition supports other options as well)." Not sure what all this means, does it mean RubberBand *can* do good resampling via SRC? Puzzled... Tim. > > Regards, > Robert > ------------------------------------------------------------------------------ Precog is a next-generation analytics platform capable of advanced analytics on semi-structured data. The platform includes APIs for building apps and a phenomenal toolset for data science. Developers can use our toolset for easy data analysis & visualization. Get a free account! http://www2.precog.com/precogplatform/slashdotnewsletter _______________________________________________ Lmuse-developer mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/lmuse-developer
