On Tue, 2011-06-07 at 15:33 +0300, Lasse K?rkk?inen wrote: > Hi, > > I tried to get as low latency as possible but there is a very > significant delay. Could someone please have a look at this simple (and > hacky) audio loopback test and point out if it could be changed to run > at lower latency? Do not mind the shortcuts taken or the obvious > distortion that occurs because capture and playback aren't synchronized > - this is simply a minimalistic latency test. > > http://codepad.org/nKNT4SyG > > Another similar test application written directly for ALSA runs with > very low latency on the same system (with pasuspender), so it is not a > hardware or driver issue. For the record, I am using generic USB > microphone for capture and HDA-Intel for playback. The buffer attributes aren't right. Set fragsize to 0 and maxlength to -1. If you set fragsize to -1 like you do now, you get the default latency for capture, which is probably pretty large. Playback latency can be improved a lot by using pa_stream_begin_write(). I don't know the exact reason why the minimum latency is so large (more than 100ms) when you don't use that. -- Tanu
