New warnings in v3.3.1 acceptable?

[david-shreffler]

Finally got v3.3.1 installed by pinning pytorch==2.3.1 and torchaudio==2.3.1 (#967 (comment)).

Transcripts are generated with diarization, but can anyone confirm with this new version that none of these (especially the last one) are an indication of whisperX not performing as expected?

New:

INFO:speechbrain.utils.quirks:Applied quirks (see `speechbrain.utils.quirks`): [allow_tf32, disable_jit_profiling]
INFO:speechbrain.utils.quirks:Excluded quirks specified by the `SB_DISABLE_QUIRKS` environment (comma-separated list): []

Existing in v3.1.1, assuming still okay?

Lightning automatically upgraded your loaded checkpoint from v1.5.4 to v2.5.0.post0. To apply the upgrade to your files permanently, run `python -m pytorch_lightning.utilities.upgrade_checkpoint whisperx/lib/python3.12/site-packages/whisperx/assets/pytorch_model.bin`
Model was trained with pyannote.audio 0.0.1, yours is 3.3.2. Bad things might happen unless you revert pyannote.audio to 0.x.
Model was trained with torch 1.10.0+cu102, yours is 2.3.1+cu118. Bad things might happen unless you revert torch to 1.x.

New:

/home/ubuntu/whisperx/lib/python3.12/site-packages/pyannote/audio/utils/reproducibility.py:74: ReproducibilityWarning: TensorFloat-32 (TF32) has been disabled as it might lead to reproducibility issues and lower accuracy.
It can be re-enabled by calling
   >>> import torch
   >>> torch.backends.cuda.matmul.allow_tf32 = True
   >>> torch.backends.cudnn.allow_tf32 = True
See https://github.com/pyannote/pyannote-audio/issues/1370 for more details.

New:

/home/ubuntu/whisperx/lib/python3.12/site-packages/pyannote/audio/models/blocks/pooling.py:104: UserWarning: std(): degrees of freedom is <= 0. Correction should be strictly less than the reduction factor (input numel divided by output numel). (Triggered internally at ../aten/src/ATen/native/ReduceOps.cpp:1807.)
  std = sequences.std(dim=-1, correction=1)

Lastly, can any of these be suppressed?

[pramadikae]

same issues with me

[Charlie11-Model]

Hi, I receive the same feedback. But mine exits with an error messageAn error occurred during diarization: list indices must be integers or slices, not tupleJust after your second last share:See https://github.com/pyannote/pyannote-audio/issues/1370 for more details.Can you help me to get the transcript with diarization successfully?

[pramadikae]

is there a solution to this issue? @m-bain @Barabazs

[591280714]

Same thing here

[kevdawg94]

+1

[cczw2010]

+1

[niradar]

+1

[danhurtes]

+1

[GlitchOwl]

+1

[weiyuwang2024]

same issue here, any solutions?

[xandark]

Apparently, no, no solutions 🫤 I've tried to pin torch and torchaudio to 2.3.1 but still cannot get this to run successfully under Ubuntu 24.04 x86_64 using uv with Python 3.10. I'm giving up.

[prkumar112451]

Hi Guys, this has nothing to do with whisperx. without whisperx, directly using diarization model of pyannote v3.1 gives this warning

/usr/local/lib/python3.11/dist-packages/pyannote/audio/models/blocks/pooling.py:104: UserWarning: std(): degrees of
freedom is <= 0. Correction should be strictly less than the reduction factor (input numel divided by output
numel). (Triggered internally at ../aten/src/ATen/native/ReduceOps.cpp:1808.)