0
0

Hi ! Ive played with the dsp_gain sample in order to integer Soundtouch to FMod. The pitch scaling function worked rather well (a little bit cracky, but still acceptable).

But the timestretch function can’t work well. I noticed that the FMOD DSP functions needs a fixed input & output buffer size, but the soundtouch procedure (putSamples(…)) returns a completly different amount of data (and also needs a different amount of input data, depending the tempo :). Is there a way to FMod to specify the input & output size in a DSP (or in a custom stream processing function, or something else ? )

  • You must to post comments
0
0

It crackles because soundtouch pitchisft is not perfectly accurate, if i put a buffer of size xxx, it will not return to me the same sized buffer (even if it’s supposed to). The author himselfs says that in it’s function comments. In fact, soundtouch isn’t a true pitch shift library, when i do pitchshift, soundtouch internally timestretch the sample and then change the frequency (!), that’s probably why there is some kind of innacuracy. Anyway, i think i can reduce crackle by tricking with soundtouch parameters.

Ive tried to use pitch shift + fmod frequency changing but there is a problem, soundtouch needs time to transform the sound. Some sample are not pitch shifted, so when i change the speed, i can heard the trick, it’s not good.

  • You must to post comments
Showing 1 result
Your Answer

Please first to submit.