Sound question

[johnl3d]

If there a program or method to extract the voice track or frequencies from say and mp3 or wav file audacity can remove most of voice but cannot fine the reverse ..yes I playing with amplitude again

[Rodney]

audacity can remove most of voice but cannot fine the reverse

 

John,

Not sure what you mean by 'cannot fine the reverse'.

[robcat2075]

In theory... if you can remove the voice and leave the music you could take that result, invert it, add it to the original and that would cancel out the music and leave the voice.

[Rodney]

In theory... if you can remove the voice and leave the music you could take that result, invert it, add it to the original and that would cancel out the music and leave the voice.

 

I see what you are saying here and can see two basic approaches. It seems that the software's (perfected) approach would eliminate the middle steps to where no inversion and cancellation is needed. I suppose the holy grail that folks would pay money for would yield the ideal process: Isolate Voice from Music. Use Voice or Music.

 

Most music production these days has all the various elements separated from the very beginning. These elements are then combined or redistro'd (as necessary) at the end of the process. Here I'm mostly thinking of right/left stereo channels and such but I sense this also provides a foundational support for the argument that layering and compositing be built into any system/production from the very start of a process and that data be maintained/secured throughout the product life cycle. Of course there can be considerably less incentive for a producer to share all of this with a consumer who might go on to alter the product.

 

I'm straying from John's topic so I'll stop that line of thinking but am just noting for myself that it is best to start from a pure source and it won't help us much if we don't have access to that source. We can get close via recreating/reverse engineering but that will always have a margin of error built in with associated costs.

 

I'm not sure of any freeware program that will effectively isolate voice from music automatically.

The closest I've come is with Audacity, manually editing elements out.

The approach I took was to go in and isolate the elements I wanted and (quickly!) deleting those I did not.

It was a very tedious process.

 

I recall once a very long time ago going through a video tutorial and editing out all of the authors 'ums' on the audio track. I also edited the voice track to have him say things that he actually did not say (he'd left out what I thought were a few key words at some stages so I found where he said those words elsewhere in the audio and inserted them where I felt they belonged... I confess that I felt a bit sinister in that process). Looking at the waveforms and starting to see similar patterns helped a lot with that. It would have certainly helped to have modern day software. I never finished that project and even thinking about doing something similar wears me out!

[robcat2075]

A quick test... importing a track, copying it to a second track, inverting one and playing them back together will get complete cancellation, total silence.

 

If you could find a track that Audacity can completely remove a vocal from without without removing music then you might be able to use this to extract a vocal only but I suspect there's always some music lost in the vocal removal process.

[jimd]

the newest version of Audacity has a vocal remove plugin ( there is also a Karaoke plugin to be had online for Audacity too )

which does a good job of removing the center vocal however ( which is basically inverting one side )

it will not remove the left and right fx vocals echos, reverb, ect..

 

 

j

[Rodney]

Thanks for that update on Audacity Jim and Robert. I thought Audacity had this cancelling capability but didn't explore the recent releases far enough to know.

I did note the karaoke option as it seemed it could be shoehorned for use with animation or in authoring tutorials (my first thought was to use it as a teleprompter).

 

We should also note that there are programs out there that specifically designed for extracting vocal elements. They range in price from hundreds to thousands of dollars for a give suite of tools. An example is Sony's recent release of SpectraLayes Pro that sells for (just short of) $400:

 

Here's a short write up on their extraction methodology:

http://www.sonycreativesoftware.com/spectr...audioextraction