đ Distil-Whisper: Distil-Large-v3.5 for CTranslate2
This repository provides the model weights of distil-large-v3.5 converted to CTranslate2 format. CTranslate2 is a high - speed inference engine for Transformer models and serves as the supported backend for the Faster-Whisper package.
đ Quick Start
This repository contains the model weights for distil-large-v3.5 converted to CTranslate2 format. CTranslate2 is a fast inference engine for Transformer models and is the supported backend for the Faster-Whisper package.
⨠Features
- Converted model weights of distil-large-v3.5 to CTranslate2 format.
- Compatible with the Faster-Whisper package for efficient automatic speech recognition.
đĻ Installation
To use the model in Faster-Whisper, first install the PyPi package according to the official instructions.
For this example, we'll also install đ¤ Datasets to load a toy audio dataset from the Hugging Face Hub:
pip install --upgrade pip
pip install --upgrade git+https://github.com/SYSTRAN/faster-whisper datasets[audio]
đģ Usage Examples
Basic Usage
import torch
from faster_whisper import WhisperModel
from datasets import load_dataset
device = "cuda" if torch.cuda.is_available() else "cpu"
compute_type = "float16" if torch.cuda.is_available() else "float32"
model = WhisperModel("distil-whisper/distil-large-v3.5-ct2", device=device, compute_type=compute_type)
dataset = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
sample = dataset[1]["audio"]["path"]
segments, info = model.transcribe(sample, beam_size=5, language="en")
for segment in segments:
print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))
Advanced Usage
To transcribe a local audio file, simply pass the path to the audio file as the audio
argument to transcribe:
segments, info = model.transcribe("audio.mp3", beam_size=5, language="en")
đ Documentation
For more information about the Distil-Large-v3.5 model, refer to the original model card.
đ License
Distil-Whisper inherits the MIT license from OpenAI's Whisper model.
đ Citation
If you use this model, please consider citing the Distil-Whisper paper:
@misc{gandhi2023distilwhisper,
title={Distil-Whisper: Robust Knowledge Distillation via Large-Scale Pseudo Labelling},
author={Sanchit Gandhi and Patrick von Platen and Alexander M. Rush},
year={2023},
eprint={2311.00430},
archivePrefix={arXiv},
primaryClass={cs.CL}
}