Pulling out just the vocals from a stereo file...FBI-style

[vanderlae]

We've all seen it in the movies...the cops are trying to hone in on the one-armed-man's location by listening to the audio from a phone-booth call. "get rid of the train..." do this, do that, and hey presto! We're left with just the voice.

Well I'm wanting to do the same thing myself. Well, sorta. I've come across a rough recording of my late grandfather singing an old country tune (cue Johhny Cash), it was clearly done on a tape recorder but his voice has this great charm...he was a wonderful singer, but...pity about his backing players. Luckily there's only an acoustic guitar and background keys fighting with the vocal, so it's not like I'm up against drums.

So I figure it'd be a nice thing to do to pluck out his vocal and re-record the instruments myself, building the track up around him. It will be an eerie, but extremely rewarding project and I'm sure my folks would like to hear him again. It's like the Beatles' lost tapes, only it's my Gramps.

The obvious thing is to cut immediately before and after each sung line, but of course you still get the guitar under it. I could low-pass to try and minimise the guitar's volume, but there's not a lot of bass coming through anyway and I'd like to retain as much of the full range of his voice as possible.

I don't have magic FBI tools, just pro-tools. So I'd like to try and do as much as I can with what I have in front of me and that includes standard digi and waves plugins (ssl, api, renaissance etc).

Does anyone have any suggestions that might help?

Thanks

[Chinagraf]

Spectral editing is probably your best bet. Izotope RX or something like that. Sometimes it can work wonders, sometimes not you just need to try it on the source. Can't remember how functional the demo version is.. I think you can't save but maybe you do something with it?
There is a free thing called photosounder that coverts a sound file to a spectral picture that you can edit in photoshop then reconvert to a sound file, but you can't hear what it is doing in real time.

[Hud]

I have a little experience with Spectral Layers, pretty powerful program (although nothing's perfect). Check out the trial, I don't remember if that's fully functional, or send me a digital file I'd be happy to have a go at it!

[Alastair Reynolds]

CDA-ProAudio are the local agent for Cedar and have devices that can do that.They sell a lot of then to Federal and State Law enforcement agencies .Give them a call,they may be able to process it as a "demo" for you or direct you to someone who can do it properly.Well worth it given it is your Grandfather and such a project is more personally important.

[vanderlae]

Thanks for the suggestions, I didn't realise there was actual technology allowing you to do this. Mighty big rock I've been living under...!

I might download a trial of one of those suggestions and see how successful I am with it.

I've always thought that editing in DAWs was akin to working graphically anyway, so doing it spectrally via photoshop makes total sense...even if it belongs on the set of CSA rather than an audio suite!

This will be fun, if it works...

[lonearranger]

Saw an article about these guys a few years ago - looks like its a send them stuff and they send it back type of approach. Would be interested to know what they quote you for this sort of job

http://www.audionamix.com/services/index.php#1

Cheers

[Dallas]

Jumping on the train real late here... would Melodyne DNA let you do this with its polyphonic editing?? (delete the instrumentation leaving the vocal line)