d EDL ASR Toolkit
A toolkit to perform online deep speech recognition
This Toolkit contains:
- A modular rule based formater deformater
- Some phonetization tools
- An audio split-on-silence tool
- A phonetic alignment tool
- An acoustic model
- A linguistic model
- An ASR engine
The environment is managed with poetry
git clone edl_asr_toolkit
cd edl_asr_toolkit
poetry install
The modules need some assets and data to work properly (tokenizers, models, etc...) You will need to provide an assets directory and a data directory with expected models, tokenizers, data etc... These are available on S3
jupyter lab
# Open notebooks in notebook folders
TBD
- Codebase linted with flake8
- Codebase autofixed with black
- Codebase tested with unittest, each submodule is tested separately