Split two voices from one file

[aa550]

I have a single file with one loud voice and a second voice in the background. How do I separate the two voices into individual files. Please help, file very important.

[alatham]

Short answer: You can't.

Long Answer: You can't, but you might be able to use the Compressor to bring that background voice closer to the front. It's impossible to say if that will work without hearing it though. But I can give you some Compressor settings to try out:

First, highlight the whole signal and use Effect -> Amplify and press ok with the default setting, the file will now be as loud as it can without clipping. This is important for the next step.

Now, highlight the whole thing again and open up Effect -> Compressor. Keep the Attack time as low as it will go, and leave the "Normalize" button checked. Try a Ratio of 5:1 and adjust the Threshold to the left until it sounds as loud as you want it. Higher Ratios will reduce the difference between quiet and loud even more.

The compressor is not perfect and you might never get results that are good enough if the difference between quiet and loud is too great. Or if the voices are talking over one another.

[aa550]

Thank you!

[kozikowski]

<<<Short answer: You can't.>>>

I'm with him. You can change it, but you might not be able to make it better. If the quieter voice was about the same volume as some of the room noises, then the voice and noise are going to come up together by using the volume compressor tool.

You can even get exotic. As long as the two never talk at the same time, manually edit one out of the other (making extensive use of the waveform tools) and change the volumes and then try to marry them again. I'd be more comfortable doing that in a different package, but I suppose it could be done in Audacity.

It's worth a shot. As I understand it, right now you have no show.

Koz

[aa550]

Thanks Koz

[soundfreak55]

Use a spectrum editor, like Jaco from http://www.soundboot.com. It can be done if the sound components are not too dense or heavily overlapping.

[steve]

Use a spectrum editor, like Jaco from http://www.soundboot.com. It can be done if the sound components are not too dense or heavily overlapping.

"Jaco" is not available at that web site. It appears that they are offering a commercial service "You send us your sound file and we take care of it!" but there is nothing on that site to help you "split two voices from one file".

Also, referring to the original question:

I have a single file with one loud voice and a second voice in the background. How do I separate the two voices into individual files.

The task here will involve waveforms that are in the same frequency band, so the techniques being used by soundboot will not work very well as they depend on spectrum separation (which is not the case with two voices).

There are tools of the same type as used by Soundboot available:
http://www.kvraudio.com/get/1651.html
http://www.cloneensemble.com/vt_main.htm

but your mileage is likely to be limited.