Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Token-level timestamps reset at breakpoints when multi-threading #2702

Open
lreardon opened this issue Jan 3, 2025 · 0 comments
Open

Token-level timestamps reset at breakpoints when multi-threading #2702

lreardon opened this issue Jan 3, 2025 · 0 comments

Comments

@lreardon
Copy link

lreardon commented Jan 3, 2025

My invocation of the command is as follows:

$PROJECT_ROOT/build/bin/whisper-cli -m $PROJECT_ROOT/models/ggml-$model_name.bin -f audio/$audio_name/preprocessed.wav -ojf -of outputs/$audio_name.transcribed -p X

-ojf outputs to json "full" (i.e. with more data).
X is the number of threads to use for transcription.

When X > 1, then at the file breakpoints, the token-level timestamps reset to 0. However, the phrase-level timestamps remain accurate.

A sample output is attached, where X = 4:
lin-alg-lect.transcribed.json

@lreardon lreardon changed the title Word-by-word timestamps restart when multithreading Token-level timestamps reset at breakpoints when multithreading Jan 3, 2025
@lreardon lreardon changed the title Token-level timestamps reset at breakpoints when multithreading Token-level timestamps reset at breakpoints when multi-threading Jan 3, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant