Drop in a clip. It separates the dialogue from the music and effects, measures how far under it was sitting, and gives it back audible. Up to 120 seconds — and it is slow, because it runs on a small CPU. Reckon on about two minutes for every twenty-five seconds of clip.
Runs on one small machine, one clip at a time, so a queue is
normal. Nothing is uploaded anywhere else. Separation model is
BandIt
(weights CC BY-NC 4.0 — this is not a commercial service).
This is not a hearing aid and it does not diagnose anything.