[linux-audio-user] Removing voice

[Kai Vehmanen]

On Wed, 30 Mar 2005, Bill Moseley wrote:

> Thanks Steve.  I'll try out ecasound.  Any suggestions on tuning
> Karaoke?  I couldn't get as good results as I could with xmms'
> plugins.

I'm no expert at this, but few tips:

sh> ecasound -i foo.wav -o alsa -el:karaoke,-3
ecasound> start
ecasound> cop-set 1,1,XXX

... replace XXX with parameters to karaoke plugin. You can hear the result 
in real-time (replace 'alsa' with 'jack_alsa' or '/dev/dsp' if needed).

You can also add multiple effects. For example to add a separate bass 
chain (not processed with karaoke plugin):

sh> ecasound -a:1,2 -i foo.wav -o alsa -a:1 -el:karaoke,-40 -ea:200 -a:2 -efl:400 -c
ecasound> start

That mixes together karaoke-processed signal (with -40dB reduction to 
vocals), with a separate chain for bass (<400Hz). You should play with the 
parameters to get best results. You can of course use other LADSPA plugins 
instead -efl (ecasound's internal lowpass filter), and also add more 
effects. To control karaoke parameters with the above setup:

ecasound> c-select 1
ecasound> co-set 1,1,XXX

Another useful command is "c-bypass" which turns of all the effects. Issue 
"c-bypass" again to reactive the plugins. You need to use "c-select" for 
selecting which processing chain to bypass.

Once you've found good set of parameters, omit the "-c" option (for 
interactive mode) and replace "alsa" with a filename, and you are ready to 
batch-process files.

More info about the available interactive commands can be found from "man 
ecasound-iam" and "man ecasound".

PS If someone can come up with good sets of parameters, let me know
    and I'll add them as default presets to ecasound for everyone
    to use. You can make a preset out of the above by simply adding the
    following to your ~/ecasound/effect_presets:

--cut--
voiceremove = -el:karaoke,-40 -ea:150 | -efl:400
--cut--

The '|' separates parallel processing chains. You can then use the preset
by issuing:

sh> ecasound -i foo.mp3 -o alsa -pn:voiceremove -c

... you can also parametrize it:

--cut--
voiceremove_varcutoff = -el:karaoke,-40 -ea:150 | -efl:%1
--cut--

sh> ecasound -i foo.mp3 -o alsa -pn:voiceremove_varcutoff,500 -c
ecasound> start
ecasound> cop-set 1,1,600

... and so on. :)

-- 
  http://www.eca.cx
  Audio software for Linux!