Speech recognition
Written by
Updated at September 24, 2024
Getting started
-
Install Docker Engine
and Docker Compose on your server. -
Get authenticated in Container Registry.
For the recommended hardware configuration, see System requirements.
Installing and setting up the streaming speech recognition service
-
Put the
docker-compose.yaml
file you got from the SpeechKit team in a folder on your server. -
Edit
docker-compose.yaml
: replaceN
with your number of CPU threads in theCPU_THREADS_NUM: N
parameter.In Linux, you can find out the number of threads on your server using the following command:
cat /proc/cpuinfo | grep -E "processor.+:" | wc -l
-
Download all the required Docker containers:
docker-compose pull
-
Start SpeechKit:
docker-compose up -d
-
Wait until the service loads
docker logs stt-server
stt-server | Server listening on 0.0.0.0:50053 stt-server | Waiting for join...
The speech recognition service will be available on TCP port 8080.