-
The problem seems to arise in assign_word_speakers() in whisperx.diarize.
The error I got was List indices can be integers or slices, got strings.
-
The results of "Audio to text" with Whisper, are textually impressive with Subtitle Edit 3.6.10 NEXT beta 219, but the synchronisation between spoken words and subtitles, is not as good as it should b…
-
Hi, are your build instructions distro agnostic? I'm not sure what I'm doing wrong but I can't build any image on Fedora 38.
Fedora usually has really up to date packages, so I'm not sure what is t…
-
about the two new added parameters in v0.8.0
I assume these are used to handle the repetition issue , I am glad to see it
the question is what are the suitable values for it
`repetition_pena…
-
`[/usr/local/lib/python3.10/dist-packages/whisperx/asr.py](https://zodbzjljtq-496ff2e9c6d22116-0-colab.googleusercontent.com/outputframe.html?vrz=colab_20230831-060138_RC00_561609135#) in load_model(w…
-
Hi,
I'm using whisperX with the `--threads 8` parameter.
For the transcription phase seems to be using only 4 threads, while for alignment and diarisation, it uses 8.
Is this the intended beh…
-
Hello
I have a strange behavior.
If a word is followed by silence, the end of the word is far from his start
EX: a 1s approx. word got a duration of 4s approx.
I think is an issue with the seg…
-
vad_options={"vad_onset": vad_onset, "vad_offset": vad_offset},Hello, the test parameters are not effective, please help to check。
model = whisperx.load_model("/data/model/"+model_path, device, compu…
-
I am loading the alignment model for CS like this:
```
alignment_model_cs,metadata_cs=whisperx.load_align_model(
language_code="cs", device="cuda:0")
```
it is giving me the followi…
-