if you want to try it out and make changes, I recommend you cache the Speech to Text binaries locally first;

./dl.sh

docker build -t stt .

docker run -d --rm -p 82:80 --name stt -i stt

post data to service for testing (this can take mp3s or other type files as input, sample shown with wav)

curl -X POST -F "file=@sample/1284-1180-0010.wav" http://localhost:82/api/v1/stt

should return JSON

{"text": "uncknockeatthedoor of the house and a chubby pleasant faced woman dressed all the blue opened it and greeted the visitors with a smile"}

Gunicorn has a timeout of 120 seconds (see last line of Dockerfile). CPU processing usually gives a 1x performance with the duration of the file, GPU performance is estimated at 0.4x the time of the file. You will need to increase this timeout for larger files to work.