22 Feb
2015
22 Feb
'15
9:34 p.m.
Am 22.02.2015 um 21:16 schrieb Jim Dummett:
This is really useful. But as far as I can see, this
is for sample rate conversion, rather than time-stretching. Difference is the
time-stretching preserves the original pitch of the sound which makes it a much more
complicated conversion and more prone to unpleasant artefacts. Or I may have
mis-understood what this site is for...
First of all - DCP-o-matic does not apply any time-compression/expansion currently.
Timecompression/expansion can use different algorithms. For musical applications,
phase-vocoders and more elaborate formant-preserving methods are state-of-the art.
But in this case of media conversions, we are talking about rather small changes - nobody
speeds up or slows down a video by 40% expecting to preserve video and audio timing and
tonality. For this type of media conversion, the simpler SRC/interpolation and filtering
algorithms are sufficient in my opinion, even for time-compression and -expansion.
- Carsten