Hi list,
In a installation I'm building, I use veejay (veejayhq.net) for video display, and route the audio of the video for further mixing in pd. This works very well.
Under some circumstances, I change the video playback speed. Veejay has a feature called "trickplay" that adjusts the audio to the video length by changing the pitch much like a 33rpm vynil record played at 45rpm.
What I want is keep the original pitch of the sound. So I figured out timestretching is what to do. I can send the speed factor of the video to pd. So I tried the timestretch~ external, the vocoder example (got this advice from list archive). They both sound terrible, even with the audio files from the examples (bell.aiff).
So is it something doable in real-time or do I have to look for a workaround ? Any advice is welcome.
Charlot
On 21/09/12 08:05, Charles Goyard wrote:
Hi list,
In a installation I'm building, I use veejay (veejayhq.net) for video display, and route the audio of the video for further mixing in pd. This works very well.
Under some circumstances, I change the video playback speed. Veejay has a feature called "trickplay" that adjusts the audio to the video length by changing the pitch much like a 33rpm vynil record played at 45rpm.
What I want is keep the original pitch of the sound. So I figured out timestretching is what to do. I can send the speed factor of the video to pd. So I tried the timestretch~ external, the vocoder example (got this advice from list archive). They both sound terrible, even with the audio files from the examples (bell.aiff).
So is it something doable in real-time or do I have to look for a workaround ? Any advice is welcome.
Real-time timestretch is a hard problem. (Timestretch itself is hard).
I'd see how mixxx does it (possible wrong number of x there, but it's a free DJ-style software with sources available last time I checked). Then see if it's time stretch suits your needs and rip it out into a Pd external - possibly with luck mixxx just uses a library to do the dirty work so it might be a quick job.
Claude Heiland-Allen wrote:
Real-time timestretch is a hard problem. (Timestretch itself is hard).
Yes, sure.
I'd see how mixxx does it (possible wrong number of x there, but it's a free DJ-style software with sources available last time I checked).
Thanks for the hint (nice mixer BTW), but no, it's not good.
Someone (thanks) mentionned soundtouch~ (http://www.katjaas.nl/pitchshift/soundtouch~.html) off-list, haven't tried that yet. I'll keep the list updated.
Charlot