Subject: Re: [ecasound] delta 1010,ecasound and recording to wav
From: Julian Dobson (juliand_AT_braverock.com)
Date: Wed Oct 29 2003 - 16:03:53 EET
You might try compressing to mp3 on the fly (as you orginally stated)
but use a higher bit rate (eg 320kb/s). This will use significantly
lower CPU - I currently encode to 64 and 128 kb/s concurrently and the
64 kb/s stream takes about 40% CPU, while the 128 is about 18% (this is
on a dual PIII 450MHz machine which is pretty tame hardware by today's
standards).
You mentioned that you are doing post processing on the files,
specifically converting them to rm format. I've never tried this, but
you might be able to convert straight to rm format. To do that, change
the ecasoundrc file so that the ext-cmd-mp3-output option points to your
rm encoder, then change your ecasound command to output to mp3.
Julian
H. T Hind wrote:
> On Wed, Oct 29, 2003 at 12:59:31AM +0200, Kai Vehmanen wrote:
>
>>On Mon, 27 Oct 2003, H. T. Hind wrote:
>>
>>
>>>I'm able to record using the above samples directly to mp3 files
>>>on at least 3 channels. As someone suggested on this list,
>>>encoding to mp3 on the fly chews up a lot of CPU which might
>>>result in Xruns. To avoid that, I'm trying to record to a wav
>>>file at a lower bitrate.
>>
>>The CPU load is only one source for xruns. In theory (and on well-tuned
>>machines, even in practise) you can do audio processing without xruns even
>>with very high CPU loads. You can get pretty far just by running ecasound
>>as root and specifying "-r" (and jackd too (-R) if you are using it).
>
>
> In production, I'd like the ability to record 8 channels
> simultaneously , each recording to individual files.
> I will try ecasound with -r option with 4 channels.
>
>
>>If you still want to try lowering the CPU load, you'll get the biggest
>>savings when you move from mp3 output to wav output. From there, lowering
>>the quality of written wav files does not help as much. Anyways, you can
>>still do it:
>
>
> The lowering is reduce disk space. In the wav format ecasound
> hardly takes up CPU time. The default format of 32,1,44100
> generates huge files and I want to be able record 24/7 if
> necessary.
>
>
>>>ecasound -f:32,1,44100 -G:jack,ecasound,recv -a:1 -i jack -o
>>>~/1.wav -a:2 -i jack -o ~/2.wav -a:3 -i jack -o
>>>~/3.wav -a:4 -i jack -o ~/4.wav
>
>
>
>>Replace -f:32,1,44100 with -f:16,1,44100 is one example.
>
>
> Often this produces big gaps (silence) in the beginning of the files.
> Any idea how I can fix that ?
>
>
>>>Is there a way to get ecasound to start writing to another file
>>>while still recording ? Ideally, I'd like to have the ecasound
>>>process start recording and then write to another file(s) at a
>>>configurable amount of time. Is it possible ?
>>
>>Hmm, so do you want to write the same material to multiple files, or do
>>you want to discard the start of the recorded material, or only write
>>to a file for a certain period of time? Ecasound can do all these things,
>>but the syntax varies from case to case.
>
>
> As the files need to be post processed and to be
> converted to a streaming format, I wanted the ability to write
> the files in say 1 hour chunks, so post processing can happen
> alongside. I can't directly change into streaming format as I
> need either wav or mp3 for post processing.
>
> If I could downsample to wav format while still conserving disk
> space, it would be idea. I can then compress the audio to rm and
> delete the wav files or mp3 files.
>
This archive was generated by hypermail 2b28 : Wed Oct 29 2003 - 16:02:13 EET