Align audio and text using wav2vec2-bert models

Community

• Discord: https://bit.ly/discord-uds
• Speech Recognition: https://t.me/speech_recognition_uk
• Speech Synthesis: https://t.me/speech_synthesis_uk

Install

uv venv --python 3.12

source .venv/bin/activate

uv pip install -r requirements.txt

# in development mode
uv pip install -r requirements-dev.txt

# check/format the code
ruff check --select I --fix
ruff format

Test

Download & convert the data

wget "https://github.com/egorsmkv/cv10-uk-testset-clean/releases/download/v1.0/filtered-cv10-test.zip"

unzip filtered-cv10-test.zip

wget https://raw.githubusercontent.com/egorsmkv/cv10-uk-testset-clean/refs/heads/main/rows.csv

Run aligment

python alignment.py