Setup Whisper

Overview

Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multitask model that can perform multilingual speech recognition as well as speech translation and language identification.

Important Docker Files

.env file

.env

NAME=whisper
DOMAIN=
MODEL=turbo #medium.en
ENGINE=openai_whisper
PORT=9000

docker-compose.yml file

docker-compose.yml

services:
  whisperasr:
      image: onerahmet/openai-whisper-asr-webservice:latest-gpu
      restart: unless-stopped
      container_name: ${NAME}
      environment:
        ASR_MODEL: ${MODEL}
        ASR_ENGINE: ${ENGINE}
      networks:
        - proxy
      dns: 
        - '10.1.10.1'
      deploy:
        resources:
          reservations:
            devices:
              - driver: nvidia
                count: 1
                capabilities: [gpu]
      labels:
        swag: enable
        swag_address: ${NAME}
        swag_port: ${PORT}
        swag_proto: http
        swag_url: ${NAME}.${DOMAIN}
        com.centurylinklabs.watchtower.enable: true
networks:
  proxy:
    external: true

Start the Container

Start the Container

docker compose up -d