huggingface / distil-whisper

Distilled variant of Whisper for speech recognition. 6x faster, 50% smaller, within 1% word error rate.

huggingface/distil-whisper Issues

Unable to set concatenate_audio parameter to False in run_pseudo_labelling.py
Closed 11 days ago
Problems in concatenate_dataset
Closed 11 days ago
ZeroDivisionError: division by zero
Updated 18 days ago
Repository Not Found for url
Closed 18 days ago
Discrepancy on WER benchmark result in Tedlium dataset
Updated 19 days ago1
Unable to reproduce results from the paper
Closed a month ago6
How to set the target language for examples in README?
Updated a month ago7
[Question] Can we distill for multiple langauges for distil-small-whisper
Updated a month ago3
Resuming training fails
Updated 2 months ago3
question about when to apply WER threshold filtering strategy with concatenated audio
Updated 2 months ago
Finetuning on which model?
Updated 2 months ago1
Whisper-tiny support?
Updated 2 months ago2
RuntimeError: User specified an unsupported autocast device_type 'mps'
Updated 2 months ago
perceptually faster inference through pre-completion inference of audio
Updated 3 months ago
Quantize distil-whisper?
Updated 3 months ago
Question: should the pseudo-labelling model and teacher model be the same?
Closed 3 months ago2
BetterTransformer optimization / flash_attn{_2}
Closed 3 months ago6
How to use distil-whisper-large-v3-de-kd model from HF?
Updated 3 months ago10
[Issue] latest run_pseudo_labelling.py
Updated 3 months ago
Training README datasets table: text column and id column
Closed 3 months ago4
WER Filtering takes too long?
Updated 3 months ago3
distil-small.en AttributeError
Closed 3 months ago3
Errors running `run_long_form_eval.py`
Closed 3 months ago1
Why do we need to tokenized file_id?
Closed 3 months ago1
large-v2 for english lost voice to text
Updated 3 months ago1
any executable script for running on custom data/given dataset
Closed 3 months ago1
Voxpopuli text column "raw_text" HF dataset card shows empty string.
Closed 3 months ago1
Pseudo-labelling librispeech_asr (train.360): KeyError `train-360` when not streaming.
Updated 3 months ago1
Cached English Common Voice dataset size.
Closed 4 months ago1
transcription results are inconsistent and timestamps are None type. Issue appears in the latest version of the transformers==4.38.1.
Updated 4 months ago
Loss Weight ablation experiment
Updated 4 months ago3
Cannot do pseudo-labelling due to some error
Closed 4 months ago2
支持中文吗？
Updated 4 months ago4
Best way to implement streaming application?
Updated 4 months ago
Short form evaluation WER % for Librispeech clean test
Closed 4 months ago3
Training README triggers BuildConfig ValueError
Closed 4 months ago3
Avoid preprocessing dataset again?
Closed 5 months ago1
Is there a parameter that can keep the numbers in their original language form?
Updated 5 months ago5
Very high WER % in extensive benchmark on Fleurs
Closed 5 months ago3
I want to confirm how the knowledge organization is implemented？
Updated 5 months ago1
Is there the "distil-whisper/distil-medium" model?
Updated 5 months ago1
ddp ERROR
Updated 5 months ago1
The evaluation stuck after unwrap_model during distillation training
Closed 5 months ago1
I can only use its encoder to extract audio features, right? How should I use it? Could you provide an example
Updated 5 months ago1
What are the settings used for WER calculation in the paper?
Updated 5 months ago1
How can I assign parameters for another language transcribe task?
Closed 5 months ago2
RuntimeError: Error(s) in loading state_dict for WhisperForConditionalGeneration.
Updated 5 months ago4
High inference time when using chunk size 15
Updated 6 months ago2
NotImplementedError
Updated 6 months ago1
Accelerate
Updated 6 months ago1