Le jeudi 18 février 2010 15:41:45, Cees van Egmond a écrit :
> Hello,
Hi !
> Today I tested the live input via oss for one hour steaming to a MP3
> file and an icecast 2 server. See config below:
(...)
> While comparing the input (played from mp3-player via line-in) and
> both recorded outputs the outputs are missing 4 secs over the duration
> of the stream (1 hour). So the output is 4secs shorter than the input.
> The log shows loads of "We must catchup x.xx seconds!". The CPU load
> is araound 55% according to top.
> How is this possible and can this be resolved?
The catchup message comes from the fact that liquidsoap is late. Usually it
tires to process data at real time.
However, liquidsoap should not drop data when it catches up but process more
quickly to, as said, catchup.
Additionally, when using input.oss, as in your script, the real time
processing should be disabled because oss has its own clock and should
regulate the processing time.
This can be done by adding:
set("root.sync", false)
at the beginning of your script.
Now, for the lost seconds, I am not sure this is liquidsoap's fault. For what
I know of OSS, when recording from the input, OSS should have an internal
buffer. Hence, I would not be surprised if there would be 4 seconds between
what is received by the soundcard and what OSS feeds to liquidsoap. Hence,
when you shutdown liquidsoap, you may loose the data that is still in the OSS
buffer.
Such a fact should be very likely if those 4 seconds are lost at the end of
the stream...
Romain
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Savonet-users mailing list
Savonet-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/savonet-users
