I can’t seem to find a suitable answer to my endeavour, I thought I might ask for advice. I have a video from a conference with the speaker using english and a translater some indian language. I’m looking for a way to cut out the translator’s part entirely, audio as well as video. Is there a software that would allow to recognize and differentiate the two voices, the speaker’s voice and the translater’s voice and automatically eliminate the translator’s voice? This would prevent spending hours, cutting out each translation passage…
That’s an interesting idea but unfortunately I am not aware of such a tool.
Perhaps in near future we’ll have such AI powered video editors but for now, you’ll have to do the manual work. Not a pretty thing to do but I see no other alternative.
I had same idea some years ago, but … there no such tools. And now there no good examples of AI software that can do it correctly. Only thing that i can suggest is mute volume when you aren’t speaking, it works good, but it hard. You need manually crop records, it can’t be done in films for example. But if you are editing short videos (for social media using some filters for example) it’s good way, can try it
Hi, I read your post about the conference video (speaker in English, consecutive interpreter in an Indian language) and wanting the interpreter’s voice cut out automatically instead of manually chopping every translated passage.
You’re right that nothing off-the-shelf does this. Voice isolation tools (ElevenLabs’ isolator, etc.) separate speech from background noise, not one speaker from another. Text-based editors like Descript/VEED let you delete a transcript paragraph, but you still have to find and mark every translated passage by hand first — same manual-cutting problem, just inside a nicer UI.
I built a tool for exactly this, called DeTranslator (detranslator.com). Answering your actual question either way:
What it does: you upload the audio or video file, the pipeline automatically detects the interpreter’s speech passages and removes them, leaving the original speaker’s voice (and the video track) intact, no manual time-stamping.
Price: based on the file’s duration, $9 per hour of input audio, $3 minimum, quoted upfront from your file’s length before you’d commit to anything. A roughly 1-hour conference recording would run about $9.
Honest status check: DeTranslator is pre-launch, there’s no live upload page or payment flow on the site yet, so I can’t hand you a link to go process this today. If you reply here (or reach out directly) with a bit about your file, I’ll keep you posted and reach out the moment the service is live, you’d be exactly the kind of case it’s built for.
If you’ve got questions about your specific file (length, audio quality, how the interpreter’s speech is mixed in) happy to answer here in the meantime.