MedicalWhisperRunner/docker-compose.yml at main · AImageLab-zip/MedicalWhisperRunner · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
services:
  api:
    build: .
    image: whisper-live:latest
    profiles: ["api"]
    command: >
      uvicorn server:app --host 0.0.0.0 --port ${API_PORT:-8000}
      --ssl-keyfile certs/key.pem --ssl-certfile certs/cert.pem
    ports:
      - "${API_PORT:-8000}:${API_PORT:-8000}"
      - "${STATUS_PORT:-8001}:${STATUS_PORT:-8001}"
    volumes:
      - ./models:/app/models:ro
      - ./certs:/app/certs:ro
    environment:
      WHISPER_MODEL: ${WHISPER_MODEL:-models/medical-whisper-large-v3-ct2}
      WHISPER_DEVICE: ${WHISPER_DEVICE:-cuda}
      WHISPER_COMPUTE: ${WHISPER_COMPUTE:-float16}
      WHISPER_DEVICE_INDEX: ${WHISPER_DEVICE_INDEX:-0,1,2}
      WHISPER_NUM_WORKERS: ${WHISPER_NUM_WORKERS:-4}
      WHISPER_API_TOKEN: ${WHISPER_API_TOKEN:?set WHISPER_API_TOKEN in the environment or a .env file}
      WHISPER_STATUS_PORT: ${STATUS_PORT:-8001}
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    restart: unless-stopped

  web:
    build: .
    image: whisper-live:latest
    profiles: ["web"]
    command: >
      uvicorn webserver:app --host 0.0.0.0 --port ${WEB_PORT:-8080}
      --ssl-keyfile certs/key.pem --ssl-certfile certs/cert.pem
    ports:
      - "${WEB_PORT:-8080}:${WEB_PORT:-8080}"
    volumes:
      - ./certs:/app/certs:ro
    environment:
      WHISPER_API_TOKEN: ${WHISPER_API_TOKEN:?set WHISPER_API_TOKEN in the environment or a .env file}
    restart: unless-stopped

  # --- proxy profile: behind the shared nginx-proxy + acme-companion stack -----
  # No self-signed cert, no directly published ports — nginx-proxy terminates TLS
  # with a real Let's Encrypt cert for PROXY_DOMAIN and forwards plain HTTP over
  # proxy-net. Both services share one hostname; nginx-proxy's VIRTUAL_PATH does
  # path-based routing so /ws and /load reach api-proxy while everything else
  # (the static page) falls through to web-proxy. See README "Running behind
  # nginx-proxy" for the caveat this implies for static/index.html's hardcoded
  # API_PORT.
  api-proxy:
    build: .
    image: whisper-live:latest
    profiles: ["proxy"]
    command: uvicorn server:app --host 0.0.0.0 --port 8000
    volumes:
      - ./models:/app/models:ro
    environment:
      WHISPER_MODEL: ${WHISPER_MODEL:-models/medical-whisper-large-v3-ct2}
      WHISPER_DEVICE: ${WHISPER_DEVICE:-cuda}
      WHISPER_COMPUTE: ${WHISPER_COMPUTE:-float16}
      WHISPER_DEVICE_INDEX: ${WHISPER_DEVICE_INDEX:-0,1,2}
      WHISPER_NUM_WORKERS: ${WHISPER_NUM_WORKERS:-4}
      WHISPER_API_TOKEN: ${WHISPER_API_TOKEN:?set WHISPER_API_TOKEN in the environment or a .env file}
      VIRTUAL_HOST: ${PROXY_DOMAIN:-}
      VIRTUAL_PORT: 8000
      VIRTUAL_PATH: /ws,/load
      LETSENCRYPT_HOST: ${PROXY_DOMAIN:-}
      LETSENCRYPT_EMAIL: ${LETSENCRYPT_EMAIL:-}
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    networks:
      - proxy-net
    restart: unless-stopped

  web-proxy:
    build: .
    image: whisper-live:latest
    profiles: ["proxy"]
    command: uvicorn webserver:app --host 0.0.0.0 --port 8080
    environment:
      WHISPER_API_TOKEN: ${WHISPER_API_TOKEN:?set WHISPER_API_TOKEN in the environment or a .env file}
      VIRTUAL_HOST: ${PROXY_DOMAIN:-}
      VIRTUAL_PORT: 8080
      LETSENCRYPT_HOST: ${PROXY_DOMAIN:-}
      LETSENCRYPT_EMAIL: ${LETSENCRYPT_EMAIL:-}
    networks:
      - proxy-net
    restart: unless-stopped

networks:
  proxy-net:
    external: true