MSTT FunASR Plugin

This repository contains the mstt-funasr plugin for the Modern Speech-to-Text (MSTT) library. This plugin provides integration with the FunASR models, allowing MSTT to leverage FunASR's capabilities for high-performance speech-to-text transcription.

Features

• Integrates FunASR models with the MSTT unified API.
• Supports various FunASR models for different languages and tasks.
• Provides efficient and accurate speech transcription.

Installation

To use this plugin, you need to have MSTT installed first. Then, you can install mstt-funasr using pip:

pip install mstt-funasr

This will also install the necessary funasr library as a dependency.

Usage

Once installed, the funasr model will be automatically available through the MSTT library. You can use it by specifying model_id="<funasr_model>" when initializing the MSTT client, <funasr_model> can be one of:

• funasr/iic/SenseVoiceSmall
• sensevoice_small
• funasr/iic/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch
• paraformer-zh
• funasr/iic/speech_paraformer-large-vad-punc_asr_nat-en-16k-common-vocab10020
• paraformer-en
• funasr/iic/speech_conformer_asr-en-16k-vocab4199-pytorch
• conformer-en
• funasr/iic/Whisper-large-v3
• whisper-large-v3
• funasr/iic/Whisper-large-v3-turbo
• whisper-large-v3-turbo

from mstt import get_model

asr_model = get_model("<funasr_model>")
result = asr_model.transcribe("path/to/your/audio.wav")

Example with CLI

You can also use the funasr model via the MSTT command-line interface:

mstt transcribe --model <funasr_model>  path/to/your/audio.wav

Contributing

Contributions to the mstt-funasr plugin are welcome! Please refer to the main MSTT repository's contributing guidelines for more information.

License

This project is licensed under the Apache License, Version 2.0 - see the LICENSE file for details.