search

tazz4843 / whisper-rs

Rust bindings to https://github.com/ggerganov/whisper.cpp
The Unlicense
607 stars 105 forks source link

Add safe bindings for speaker diarization #123

Closed WenqingZong closed 3 months ago

WenqingZong commented 3 months ago

As title

thewh1teagle commented 2 months ago

@WenqingZong Do you know if we can get speaker labels (such as numbers / ID)? I can only see that we can get speaker turns, but I didn't understood if we able to get speaker labels so we know who is talking (ID) and when

WenqingZong commented 2 months ago

@WenqingZong Do you know if we can get speaker labels (such as numbers / ID)? I can only see that we can get speaker turns, but I didn't understood if we able to get speaker labels so we know who is talking (ID) and when

Sadly, I don't know any easy ways to achieve what we want. I do agree only getting [SPEAKER_TURN] is not useful enough and more like a toy.

A relevant issue is also raised in the underlying repo.