Software to take out singing (voice) from recording

[Kennykay]

Please I should be grateful if Doom9 or any forum members would help me. I wish to find out if there is any available software that will enable me to extract out the singing from an audio recording - cd or audio cassette, leaving only the accompaniment. I intend to take out the singing voice(s) in some of my audio recordings, then be able to convert the accompaniment to midi and load the converted accompaniments onto a music arranger keyboard and play them back as backing tracks.

I know that Doom9 has been very kind to provide us with numerous softwares to aid in various areas. I'll appreciate help on this please.

Thank you.

Kennykay

[JohnnyMalaria]

Try inverting one of the channels (left or right, it doesn't matter) and then create a mono track from the two. Typically, the vocal is mixed dead center so it is in both channels to the same extent. Inverting one channel then adding them to make the mono track causes the vocal to be (almost completely) eliminated. This is the trick used to create karaoke tracks.

[Kennykay]

Try inverting one of the channels (left or right, it doesn't matter) and then create a mono track from the two. Typically, the vocal is mixed dead center so it is in both channels to the same extent. Inverting one channel then adding them to make the mono track causes the vocal to be (almost completely) eliminated. This is the trick used to create karaoke tracks.
Hi Jonny, thanks for your kind reply. I don't actually understand what you mean by 'inverting' one of the channels. If you can make it clearer to me, I will appreciate it. Also, at the bottom of your reply, you included a link to Enosoft DV Processor. may I know if this software would help in the process.

Thank you.

Kenny

[JohnnyMalaria]

Our software will help if your material is DV format. Otherwise not.

By 'invert' I mean literally turn the signal upside down just as if a speaker were to be wired the wrong way around. Basically, make all the +ve values -ve and all the -ve ones +ve. Most audio editors should be able to do this. If you imagine a truly mono source (L and R identical), then inverting say R and then combining them will lead to silence (instead of L + R you get L - R).

I've found this tutorial for achieving the effect with Audacity:

http://yudisulistya.blog.unair.ac.id/2008/07/01/remove-vocals-from-audio-using-audacity/

What format are your audio files?

[leeperry]

yeah you could play around at inverting the phase of one channel, turn the original to mono and add them together.

otherwise a few more software to remove vocals :
http://www.elevayta.net/product13.htm

http://wareseeker.com/Audio-Multimedia/analogx-vocal-remover-1.00.zip/12337

[netmask]

Please I should be grateful if Doom9 or any forum members would help me. I wish to find out if there is any available software that will enable me to extract out the singing from an audio recording - cd or audio cassette, leaving only the accompaniment. I intend to take out the singing voice(s) in some of my audio recordings, then be able to convert the accompaniment to midi and load the converted accompaniments onto a music arranger keyboard and play them back as backing tracks.

I know that Doom9 has been very kind to provide us with numerous softwares to aid in various areas. I'll appreciate help on this please.

Thank you.

Kennykay

You can remove the vocal assuming it has been placed accurately in the centre of the sound stage by inverting or putting say the left channel out of phase and then mixing the resulting left and right channel. Any information that is common to the left and right channel will be cancelled out albeit allowing for accuracies of the original mix. You can do this quite simply in CoolEdit Pro. Open the file and go to menu item Effects---->Amplitude------>channel Mixer----->Vocal Cut voila! done

Next you say you are going to convert the accompaniment sans vocal to MIDI. By that I assume you are going to sit down and manually transcribe the notes either onto manuscript or a keyboard as there simply isn't a program that will literally "convert" a complex audio file to a MIDI data set. MIDI is the equivalent of punched paper roll found in old pianolos and as such does not "contain" any audio just an instruction set to tell your synth what to play.

Caveat: If the accompaniment is a single instrument like a guitar and it is a slow piece then there are transcribing software that can cope but you usually have to do a hell of lot of editing. If it is a rock band or a symphonic orchestra, forget it the result will be a midi file set for 3 or 4 pianos and a jumble of notes. A good FAQ on this can be found here http://home.roadrunner.com/~cosmogony/ammfaq.html#2.07

I have been down this track, I'm a classical composer and audio engineer, very frustrating but I do believe it cannot be done with any degree of accuracy except for simple one note at a time situations - many have tried. A old trick some of the students at my Uni used to do in the contemporary composers group was take a well known classical overture, use one of the so called wav to midi converters and then use the resulting jumble as a basis for a "new" original work that had a strange familiarity about it but that was about all! Similar to graphic artists taking a well known image and distorting it... ART

[raquete]

you can try dsp_centercut from http://www.moitah.net
if the voice is in the center, use the sides. if the voice is in one side only, use the center and the "other" side....and so on.
work in winamp or with ffdshow(DirectX) inside audition,cooledit and others audio editors.