fujielab-asr 0.1.4

Automatic Speech Recognition (ASR) tools for Fujie Lab

fujielab-asr

Automatic Speech Recognition Modules for Fujie Laboratory based on ESPnet.

Features

• Supports streaming ASR using Contextual Block Transformer and RNN-T

Installation

Requirements

• Python >=3.10,<3.12 (later versions may not work, because of the dependency on ESPnet)
• ESPnet (202301 or later) required version is different for a model you want to use

Installation with pip

You can install the latest version of fujielab-asr from PyPI using pip:

pip install fujielab-asr

Note that you need to use Python 3.11, so you may need to create a virtual environment first.

Installation from source

In the case you want to use conda environment:

conda create -n fujielab-asr python=3.11
conda activate fujielab-asr
pip install -e .

Example Usage

Check the example scripts in the examples directory.

• examples/run_streaming_asr.py: Example of streaming ASR.
• examples/run_streaming_asr_live.py: Example of live streaming ASR.
• examples/demo.py: Gradio demo for speech recognition.