From: Willian J. G. <wil...@gm...> - 2011-02-01 17:52:46
|
Hello David... i got it the issue... but now... what's i have to do? what you think if i do this: input = mksafe(buffer(input.alsa(device="pcm.liquidsoap"), buffer=5, max=10)) thanks for your attention :) On Sun, Jan 30, 2011 at 2:48 PM, David Baelde <dav...@en...>wrote: > Hi Willian, > > The good news here is that your soundcard input is not perturbated by > what's going on. > > However I expected some buffer overrun rather than underrun. What > you're seeing means that the icecast output consumes the buffer too > fast compared to the rate at which it is being filled by your > soundcard. What's more, it happens really often. It seems to indicate > a severe drift between your computer and your soundcard's clock... > > Let's try to sum up to make sure we don't miss anything here. > * In your first version, ALSA was doing the sync of everything (note > for example the wallclock_alsa in the logs) but there were overruns. > Those can be caused if icecast takes too long to send the data -- in > the meantime, ALSA's internal buffer overflows. > * In the second version we put the icecast output in another clock, > but now the buffer between the input and output empties quite often, > showing that icecast is too fast, contradicting the previous case. > > Tell me if that makes sense; perhaps you'll realize something funny is > going on... > > In the meantime you can tweak the buffer()'s parameters, in particular > buffer and max. > > Cheers, > -- > David > |