Am Tue, 29 Jun 2010 12:54:40 +0100
schrieb zonyi besk <zo...@gm...>:
> > I tried like that and thought it's not working as theres no audio device
> > (soundcard).
> >
> > pv -L 15000 foo.mp3 | mpg123 --timeout -1 --smooth -q -R --fifo
> > /dev/shm/mpg123-cmd1 --remote-err -s --wav - | lame - > out.mp3
> > echo "SILENCE" > /dev/shm/mpg123-cmd1
> > echo "L /dev/stdin" > /dev/shm/mpg123-cmd1
> > echo "PITCH 0.8" > /dev/shm/mpg123-cmd1
Whoa. You want to create a pitched WAV? That's not what the pitch hack was intended for. It does work in such a way:
mpg123 -w pitched.wav --pitch -0.5 test.mp3
There, it simply creates a WAV file with half the sample rate as the input mp3 has, resulting in downpitch on playback.
That also works (note that -s is not needed, it actually conflicts with -w)...
mpg123 -w - --pitch 1 -R --remote-err > out.wav
@R MPG123 (ThOr) v6
silence
@silence
load file.mp3
@I low_ao_mix
@P 2
@S 1.0 3 44100 Joint-Stereo 2 104 2 0 0 0 32 0 1
@P 0
quit
> > Cannot rewind WAV file. File-format isn't fully conform now
Yes... that's always the case when writing into a pipe. The length info should be missing, that doesn't have to be fatal.
But... as I state below: This kind of writing WAV data is indeed broken, currently:-/
> its created
> > without pitch
Yeah, the pitch works via the sample rate header entry, and I realize that this might not be correct for writing to a pipe. Actually, I see that it will only default to stereo, 44100 Hz ... hm, I do realize that I should not recommend writing WAV files to stdout as it is now (mpg123 -w - file.mp3 > out.wav works, different from mpg123 -w - file.mp3 | cat > out.wav).
The code won't necessarily transport the proper header information... we might fix that in future, but for now you should better go back to plain RAW writing to standard output via -s.
> > echo "SILENCE" > /dev/shm/mpg123-cmd1
> > echo "L /dev/stdin" > /dev/shm/mpg123-cmd1
> > echo "PITCH 0.8" > /dev/shm/mpg123-cmd1
And in any case: When you want to have a fixed pitch setting from the beginning, use the --pitch command line parameter.
The code above opens the output with the native rate first, then tries to reopen to satisfy the pitch setting. That won't work with WAV output to a pipe.
Generally, changing pitch at playtime won't work at all for WAV output, as there is only one header field for a file-wide sampling rate.
Well, until the WAV writing to pipes is fixed, you will need to resort to piping via mpg123 -s ... and you then can just tell lame that the input has a different sampling rate to achieve the same pitch effect.
Yeah, either pipe through something that resamples for pitching... or tell lame a different input sample rate so that it resamples internally.
Alritghty then,
Thomas.
|
