Automatic Speech Recognition
Transformers
PyTorch
TensorFlow
JAX
Safetensors
whisper
audio
hf-asr-leaderboard
Eval Results (legacy)
Instructions to use openai/whisper-small with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use openai/whisper-small with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="openai/whisper-small")# Load model directly from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq processor = AutoProcessor.from_pretrained("openai/whisper-small") model = AutoModelForSpeechSeq2Seq.from_pretrained("openai/whisper-small") - Notebooks
- Google Colab
- Kaggle
Pred and ref are switched in the WER calculation.
#3
by lichenda - opened
Dear Whisper authors, I found that the the prediction and the reference sentences are switched in your sample code. Since the denominator in WER is the length of the ref, so the WER results will be affected. Although It shouldn't make much difference for the overall score, but the deletion and insertion errors will be switched.