lite-whisper-large-v3-turbo Open-source Speech Recognition Model - Free Deployment, Small Size and High Accuracy

Lite Whisper Large V3 Turbo

Developed by efficient-speech

Lite-Whisper is a lightweight version of OpenAI Whisper compressed using LiteASR technology, significantly reducing model size while maintaining high accuracy.

Speech Recognition

Transformers

Open Source License:Apache-2.0 #Lightweight Speech Recognition #Low-Rank Compression #Multilingual ASR

Downloads 516

Release Time : 2/26/2025

Model Overview

A lightweight version of Whisper compressed with LiteASR technology, designed for efficient automatic speech recognition tasks.

Model Features

Efficient Compression

Uses LiteASR technology to significantly reduce model size while maintaining high accuracy.

Performance Balance

Achieves a good balance between model size and recognition accuracy.

Fast Inference

Offers faster inference speed compared to the original Whisper model.

Model Capabilities

Speech-to-text

Multilingual speech recognition

Long audio processing

Use Cases

Speech Transcription

Meeting Minutes

Automatically converts meeting recordings into text transcripts

Average Word Error Rate: 12.6%

Media Caption Generation

Automatically generates subtitles for videos and podcasts

Voice Assistants

Voice Command Recognition

Implements voice command recognition on resource-constrained devices

🚀 Lite-Whisper large-v3-turbo

Lite-Whisper large-v3-turbo is a compressed version of OpenAI Whisper, offering efficient automatic speech recognition capabilities. It utilizes LiteASR technology to optimize performance.

🚀 Quick Start

This section provides an overview of the Lite-Whisper large-v3-turbo model. For detailed usage, please refer to the official Hugging Face documentation of the transformers library.

✨ Features

Compressed Model: Based on OpenAI Whisper, it uses LiteASR technology to compress the model size, reducing computational resource requirements.
High - Performance ASR: Suitable for automatic speech recognition tasks, achieving competitive results in benchmarks.

📚 Documentation

Model Information

Property	Details
Base Model	openai/whisper-large-v3-turbo
Library Name	transformers
License	apache - 2.0
Pipeline Tag	automatic-speech-recognition
Tags	audio, automatic-speech-recognition, whisper, hf-asr-leaderboard

Benchmark Results

The following table shows the average word error rate (WER) evaluated on the ESB datasets:

Model	Average WER (↓)	Encoder Size	Decoder Size
whisper-large-v3	10.1	635M	907M
lite-whisper-large-v3-acc	10.1	429M	907M
lite-whisper-large-v3	10.2	377M	907M
lite-whisper-large-v3-fast	11.3	308M	907M

whisper-large-v3-turbo	10.1	635M	172M
lite-whisper-large-v3-turbo-acc	10.2	421M	172M
lite-whisper-large-v3-turbo	12.6	374M	172M
lite-whisper-large-v3-turbo-fast	20.1	313M	172M

whisper-medium	14.8	306M	457M

Citation

If you use LiteASR in your research, please cite the following paper:

@misc{kamahori2025liteasrefficientautomaticspeech,
      title={LiteASR: Efficient Automatic Speech Recognition with Low-Rank Approximation}, 
      author={Keisuke Kamahori and Jungo Kasai and Noriyuki Kojima and Baris Kasikci},
      year={2025},
      eprint={2502.20583},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2502.20583}, 
}

📄 License

This project is licensed under the Apache 2.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご