Hi David, On Mon, 22 Oct 2001, David Burrows wrote: > My question is this. How would one go about modifying say mpg123, so that > it uses 3x256 (or 3x128) byte buffering? I've ran the source through > doxygen, and started to have a brief look into it but I am wondering if > someone has already achieved this or could give a description of where to > start.
The question becomes, why? Latency is important when receiving audio processing it, and resending it somewhere else. It is also important when composing on the computer. But playing back mp3s? What does having 2 ms buffers for mp3 playback get you? I'll have to admit I don't quite understand. > The bonus question is about pitch control. I understand that this can be > achieved by simply changing the sampling rate, however, I'm wondering if > anyone has knowledge of fast or high quality resampling algorithms? Of course, I'm biased, having worked with GStreamer a bit. We have a couple of mp3 decoders (mpg123 and mad, which is imho better), a resampler (called 'speed' - just a linear interpolator), and an alsa sink that you can set to the desired period size and number. So if you want to play around with this sort of thing, I'm sure that there are other tool kits as well. Be sure to use CVS gstreamer if at all, at least until our next release. Also, the code is on sourceforge, so you can browse it directly: gstreamer.net/cvs/ Good luck, but consider your needs first. wingo.
