tts-voice-creator-clone-and-design-2/docker-compose.yml

services:
  # The web process talks to this narrowly configured proxy over the Compose
  # network; the host Docker socket is never mounted into the web container.
  docker-socket-proxy:
    image: tecnativa/docker-socket-proxy:0.3.0
    container_name: tts-voice-creator-docker-proxy
    restart: unless-stopped
    # NOT read_only: the entrypoint sed's its own haproxy.cfg.template into
    # haproxy.cfg at startup, in the same directory the template ships in --
    # read_only crash-looped it ("Read-only file system"), and a tmpfs mount
    # over that directory only trades one failure for another by hiding the
    # template it needs to read (confirmed live: "No such file or
    # directory"). The real hardening here is cap_drop: ALL,
    # no-new-privileges, and the Docker socket itself being mounted :ro.
    cap_drop:
      - ALL
    security_opt:
      - no-new-privileges:true
    environment:
      CONTAINERS: "1"
      POST: "1"
      IMAGES: "0"
      NETWORKS: "0"
      VOLUMES: "0"
      SERVICES: "0"
      TASKS: "0"
      SWARM: "0"
      SYSTEM: "0"
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
    networks:
      - dgx_net

  tts-voice-creator-clone-and-design-2:
    build:
      context: .
      dockerfile: Dockerfile
    image: ghcr.io/martin-b78/tts-voice-creator-clone-and-design-2:${APP_TAG:-latest}
    container_name: tts-voice-creator-clone-and-design-2
    restart: unless-stopped

    # The application writes only to the explicitly mounted data paths below.
    # Keep the image filesystem immutable even if an endpoint is compromised.
    read_only: true
    cap_drop:
      - ALL
    security_opt:
      - no-new-privileges:true
    tmpfs:
      - /tmp:size=1g,mode=1777

    # Keep a runaway OCR/import/generation request from consuming all host RAM.
    # Override these per machine when large PDF jobs need more headroom:
    # APP_MEM_LIMIT=12g APP_MEM_RESERVATION=2g docker compose up -d
    mem_limit: "${APP_MEM_LIMIT:-8g}"
    mem_reservation: "${APP_MEM_RESERVATION:-2g}"

    ports:
      - "7890:7890"

    depends_on:
      - docker-socket-proxy

    volumes:
      - ${VOICE_HOST_DIR:-/home/sparky/Media/_Sounds/TTS_Voices}:/voices:rw
      - ./config:/home/app/.config/tts-voice-creator:rw   # settings, presets, routes
      - ./logs:/logs:rw                                    # application logs
      - /home/sparky/Docker/tts-dgx-spark-faster-qwen3-tts/config:/tts-config:rw
      - ./static:/app/static:ro
      - ./server.py:/app/server.py:ro
      - ./core:/app/core:ro
      - ./routes:/app/routes:ro
      - ./workflows:/app/workflows:ro
      - ./VERSION:/app/VERSION:ro
      - ./CHANGELOG.md:/app/CHANGELOG.md:ro

    environment:
      - PYTHONUNBUFFERED=1
      - VOICES_DIR=/voices
      - OUTPUT_DIR=/voices/active_voices
      # Containers that load WAV files and need a restart after voice changes
      - TTS_CONTAINER_NAME=tts-faster-qwen3-tts-voiceclone
      - TTS_CONTAINER_NAMES=tts-faster-qwen3-tts-voiceclone,tts-faster-qwen3-tts-streaming
      - VOICE_DESIGN_MODEL=Qwen3-TTS-12Hz-1.7B-VoiceDesign
      - DOCKER_HOST=tcp://docker-socket-proxy:2375
      # Word-level forced aligner sidecar; empty disables word timing (sentence timing keeps working)
      - ALIGNER_URL=${ALIGNER_URL-http://tts-aligner:8000}

    extra_hosts:
      - "host.docker.internal:host-gateway"

    networks:
      - dgx_net

  # Word-level forced aligner (torchaudio MMS_FA). Separate image so the app stays torch-free. Reachable only on
  # the Compose network (no published port); the app calls http://tts-aligner:8000. Weights (~1.2 GB) are
  # downloaded once into the named volume; after that the service needs no outbound access.
  tts-aligner:
    build:
      context: ./aligner
    image: tts-aligner:local
    container_name: tts-aligner
    restart: unless-stopped
    read_only: true
    cap_drop:
      - ALL
    security_opt:
      - no-new-privileges:true
    tmpfs:
      - /tmp:size=256m,mode=1777
    mem_limit: "${ALIGNER_MEM_LIMIT:-3g}"
    cpus: "${ALIGNER_CPUS:-4}"
    environment:
      - ALIGNER_DEVICE=cpu
      - ALIGNER_THREADS=4
    expose:
      - "8000"
    volumes:
      - aligner-models:/models:rw
    healthcheck:
      test: ["CMD", "curl", "-fsS", "http://127.0.0.1:8000/health"]
      interval: 30s
      timeout: 5s
      retries: 3
      start_period: 60s
    networks:
      - dgx_net

  # Optional GPU variant: `docker compose --profile gpu up -d --build tts-aligner-gpu` (stop tts-aligner first;
  # both answer to the name tts-aligner). Off by default because the DGX GPU is often full.
  tts-aligner-gpu:
    profiles: ["gpu"]
    build:
      context: ./aligner
      args:
        TORCH_INDEX: https://download.pytorch.org/whl/cu130
    image: tts-aligner:gpu
    container_name: tts-aligner-gpu
    restart: unless-stopped
    read_only: true
    cap_drop:
      - ALL
    security_opt:
      - no-new-privileges:true
    tmpfs:
      - /tmp:size=256m,mode=1777
    mem_limit: "${ALIGNER_MEM_LIMIT:-3g}"
    environment:
      - ALIGNER_DEVICE=cuda
    expose:
      - "8000"
    volumes:
      - aligner-models:/models:rw
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    healthcheck:
      test: ["CMD", "curl", "-fsS", "http://127.0.0.1:8000/health"]
      interval: 30s
      timeout: 5s
      retries: 3
      start_period: 60s
    networks:
      dgx_net:
        aliases:
          - tts-aligner

networks:
  dgx_net:
    external: true

volumes:
  aligner-models: