separating languages in an audio file

Sagitt

Hi,

Apologies if this is not the right forum for this - feel free to move where you think it fits best.
I was just wondering if this is possible at all - to cut translator out and leave original speaker in an audio file (mp3).
I have about 20 hours of material that I would like to listen. Original speaker speaks in English, but there is translation to another language, what doubles the time. Is there any software that you might be aware of that would be good for this job? Is this possible at all?

regards,
Sagitt

jonster

If you're looking to make a custom solution you could look at IBMs Watson combining their speech to text service with their language recognition service, timestamping occurrences of the language you wish to extract.

counterpointaud

So there is two distinct voices, never speaking at the same time? You can probably automate most of this, especially if one of the voices is band-limited. Check out Adobe Audition. You will to try few things depending on what differences the voices. Freq analyisis, noise gating. For example if one voice has content under 50hz, but the other doesn't, you can use plugins to create regions where the this content exists, and either remove them, or the other parts. I suspect you will have to listen to the full thing at some stage though, and fix anything the automated settings got wrong. It's not going to be super easy and there is no out of the box solution I know of.

Sagitt

Thank you very much for reply. I knew at the beginning that it will not be easy, just was wondering if anyone has came across similar situation. In the worst case scenario I will have to listen both languages, which is not that big deal.

regards,
sagitt