I took Vosk and wrapped it in a HTTP API.
I made a Docker image that contains everything needed but the voice models.
You can run it with:
docker run \
-v "$(pwd)/models:/app/models" \
-e PORT=13000 \
-p 13000:13000 \
victor141516/vosk-http-api
You will need to download the Vosk voice models and put them in the models
directory, having each of them in a subdirectory named after the language. e.g. models/esES
will contain the Spanish model:
TBA