At 17:34 23.08.2004, Thomas Grill wrote:
Hi Sven, Python is too slow for audio processing. However, i'm working on the "dyn" project which will allow to build networks of normal PD signal objects from within pyext. If you want even more flexibility you should a look at Supercollider.
hi thomas, I know that python is actually too slow for audio processing, but I just want to do a fft, extract critical bands and some other analysis
- feature extraction in general. all that can be done using Numeric Python
which is a highly optimized package for number crunching tasks. that package is written in c and I wouldn't get it much faster if I wrote an external in c than just using numpy. I've already checked if I could do all analysis using pd-objects and just passing the results to pyext but it doesn't seem to be possible, at least not as flexible as I'd need it. I've heard about "dyn" and the things that should get possible with it, but it doesn't help me for what i want to do. and supercollider is only for mac. if i'd just get chunks of audio into pyext it would be fine... sven.