search

miguelvalente / whisperer

Go from raw audio files to a text-audio dataset automatically with OpenAI's Whisper.
132 stars 12 forks source link

IndexError #23

Closed gnmarten closed 1 year ago

gnmarten commented 1 year ago

Very fond of this concept, thanks, massive timesaver.

However, I ran into an error. fresh_cuts has the timestamps and the speakers, but then

speaker_embeds = embed(speakers_files[0], embedder)

leads to

----> 1 speaker_embeds = embed(speakers_files[0], embedder)
IndexError: list index out of range

When are the separate "audio_files_wav/speakers" supposed to be created? Because at this stage, they are not there yet, right?

Running TestDiarize.ipynb on Colab with Python 3.8 rather than Python 3.10 (wav file is about 20 minutes, one of the speakers is a bit muffled because in a Zoom call)

Any suggestions?

miguelvalente commented 1 year ago

Thanks 😄. Can you show me what your data directory looks like? Including any files that got generated. The "speakers/" dir gets populated at the end of the diarization. Maybe i can provide a better error message.

gnmarten commented 1 year ago

sorry, my bad, I assumed the notebook contained the whole pipeline, but I think it doesn't I am now running the python commands rather than the notebook, now the directories are created in due order works like a charm, this is splendid just thinking about the hoops I had to jump through to arrive at the result earlier (transcribing twice in fact ...) I expect this repository to receive a lot of stars!

miguelvalente commented 1 year ago

Thanks! Let me know if you run into any other issues.