Implementation of an API to stream audio files for transcription in Django using the REST Framework.
If you want to try the server inside a virtual environment.
pip install -r requirements.txt
Update docker-compose.yml
to point to your models path under services:stt:volumes
.
Build the stack using docker-compose
.
docker-compose up -d
Then collect staticfiles.
docker-compose run --rm stt /app/manage.py collectstatic --no-input
Go to http://0.0.0.0/stt/
.
Before you run the server, you need to update the configuration located at stt_app/config/config.json
with the path to your desired models.
{
"stt": {
"model" :"path/to/models/output_graph.tflite",
"lm": "path/to/models/kenlm.scorer",
"audiofiledir": "/path/to/tmp_dir_audio",
"audiofilelength": "10",
"debug": "1"
}
}
Run the server in local for testing (HTTP).
python manage.py runserver
Run the server in local for production (HTTPS).
python manage.py runsslserver
This project was inspired and uses modified code from: