-
-
Notifications
You must be signed in to change notification settings - Fork 1.4k
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
Default config of without_timestamps=True
affects whisper transcript quality.
#932
Comments
yes that's right, when without_timestamps=False some of the previously untranscribed ones become transcribed by changing them to false |
where should I change this parameter without_timestamps to False in whsperx ? |
in the load_model function you can add the asr_options parameter, as in the following example: asr_options = { model = whisperx.load_model("large-v3", device="cuda", asr_options=asr_options) |
@pramadikaegamo thanks a lot . |
Can he only use this parameter in python and not on the command line? |
Yes, of course, that's what I gave, for example using parameters in Python, not in the command line |
WhisperX default is
without_timestamps=True
; while faster-whisper's default iswithout_timestamps=False
. This affects transcript quality, whisperX output can have long (5-15s) continuous transcript drops. Attached an example of a small audio clip (~10s, so VADs in WhisperX and Faster-Whisper are off) that has worse transcript output whenwithout_timestamps=True
.whisperxWithoutTimestepsExample.zip
The text was updated successfully, but these errors were encountered: