Extract Vocal From Audio.

[Jose Arrieta]

Hi guys, What are the best technique to extract vocals from Audio with background music?. If you could explain the technique too, would be great.

Thank you very much in advance. 

[Justin Salamon]

While latest doesn't always mean greatest, the MIREX 2016 singing voice separation challenge results might be a good place to start: http://www.music-ir.org/mirex/wiki/2016:Singing_Voice_Separation_Results

Off the top of my head, some open source algorithms include (but aren't limited to):

https://github.com/IoSR-Surrey/untwist

http://www.durrieu.ch/research/jstsp2010.html

http://www.zafarrafii.com/repet.html#Codes

Cheers

p.s - this might be a better question for the ISMIR community mailing list?

p.p.s - as far as librosa is concerned I think (?) the closest thing you could play around with is HPSS, I think that's been used for singing voice separation (or enhancement) in the past.

On Tue, Sep 27, 2016 at 3:20 PM, Jose Arrieta <jmarr...@gmail.com> wrote:

Hi guys, What are the best technique to extract vocals from Audio with background music?. If you could explain the technique too, would be great.

Thank you very much in advance. 

--
You received this message because you are subscribed to the Google Groups "librosa" group.
To unsubscribe from this group and stop receiving emails from it, send an email to librosa+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/librosa/10e45cf0-2feb-4419-b625-d140319101e0%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Justin Salamon, PhD

Senior Research Scientist

Music and Audio Research Laboratory (MARL)

& Center for Urban Science and Progress (CUSP)

New York University, New York, NY

www.justinsalamon.com

[Brian McFee]

Just tacking onto this: librosa makes it easy to implement repet-sim using librosa.decompose.nl_filter.

Here's an example in the gallery.  The method works well on sample-based music where the non-vocal parts are highly repetitive.  You might be able to enhance it a bit using HPSS, as Justin suggests; I think Meinard Mueller's group has a recent paper (icassp2016?) that does something similar using hpss+rpca.

On Tuesday, September 27, 2016 at 3:51:16 PM UTC-4, Justin Salamon wrote:

While latest doesn't always mean greatest, the MIREX 2016 singing voice separation challenge results might be a good place to start: http://www.music-ir.org/mirex/wiki/2016:Singing_Voice_Separation_Results

Off the top of my head, some open source algorithms include (but aren't limited to):

https://github.com/IoSR-Surrey/untwist

http://www.durrieu.ch/research/jstsp2010.html

http://www.zafarrafii.com/repet.html#Codes

Cheers

p.s - this might be a better question for the ISMIR community mailing list?

p.p.s - as far as librosa is concerned I think (?) the closest thing you could play around with is HPSS, I think that's been used for singing voice separation (or enhancement) in the past.

On Tue, Sep 27, 2016 at 3:20 PM, Jose Arrieta wrote:

Hi guys, What are the best technique to extract vocals from Audio with background music?. If you could explain the technique too, would be great.

Thank you very much in advance. 

--