Subject: Re: [ecasound] Re: ecasound latency (fwd)
From: Kai Vehmanen (firstname.lastname@example.org)
Date: Thu Jan 20 2000 - 17:39:25 EET
On Thu, 20 Jan 2000, Jeremy Hall wrote:
>> too much. For my own realtime processing needs, I usually use -b:64
>> (~1.5ms) or -b:128 (~2.9ms). These are almost inaudible.
> but 2.9 multiplied by 8 or 16 yields 23.2 or 46.44. This is NOT
> acceptable when you're trying to run 8 or 16 tracks.
In this case, number of tracks doesn't affect latency. Basicly,
ecasound engine ask for buffersize samples from every input, processes
this data (tracks * buffersize samples) and then writes to all outputs.
Realtime devices are active all the time, so ecasound has to be able
to complete the whole processing loop (input - process - output)
during (bsize/sample_rate*1000) msecs. If it doesn't, underruns will
occur on inputs, and overruns on outputs. But latency won't be
-- Kai Vehmanen <email@example.com> -------- CS, University of Turku, Finland . http://www.wakkanet.fi/ecasound/ - linux multitrack audio processing . http://www.wakkanet.fi/sculpscape/ - ambient-idm-rock-... mp3/ra/wav
This archive was generated by hypermail 2a24 : Thu Jan 20 2000 - 16:46:43 EET