Instructions to use distil-whisper/distil-large-v3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use distil-whisper/distil-large-v3 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="distil-whisper/distil-large-v3")# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq processor = AutoProcessor.from_pretrained("distil-whisper/distil-large-v3") model = AutoModelForSpeechSeq2Seq.from_pretrained("distil-whisper/distil-large-v3", device_map="auto") - Transformers.js
How to use distil-whisper/distil-large-v3 with Transformers.js:
// npm i @huggingface/transformers import { pipeline } from '@huggingface/transformers'; // Allocate pipeline const pipe = await pipeline('automatic-speech-recognition', 'distil-whisper/distil-large-v3'); - Notebooks
- Google Colab
- Kaggle
Transcribe into a different language than English...
Is it possible to transcribe to a different language than English? I'm using a pipeline instance, and when I try to use the language="pt" or language="portuguese", I get errors of not recognising the kwarg.
Hi @IvoTavares , I'm trying to get it in other languages too, but here my audio in Portuguese is transcribed to English
Hey there 🤗distil-large-v3 checkpoint in an English-distilled version of large-v3, meaning it is monolingual. There exist other languages (like French and Brazilian Portuguese), yet I would strongly recommend trying large-v3-turbo, a multilingual whisper checkpoint that should offer great speed ups from large-v3 and close to perfs compared to distil-large-v3 (distil is a 2 layers decoder while turbo is 4 layers vs large that is 32).