homeclaw/Dockerfile.voice

# ============================================================
# SmallClaw Voice – Dockerfile.voice
# ============================================================
# Voice-enabled build with ffmpeg, whisper.cpp (STT), and Piper (TTS).
# Use this when deploying with voice support on a capable server.
#
# Build:
#   docker build -f Dockerfile.voice -t smallclaw-voice .
#
# Run:
#   docker run -d --name smallclaw-voice \
#     -p 18789:18789 \
#     -v smallclaw_data:/data \
#     -v smallclaw_workspace:/data/workspace \
#     -v ~/.localclaw:/root/.localclaw \
#     -e OLLAMA_HOST=http://host.docker.internal:11434 \
#     smallclaw-voice
#
# Or with docker compose (see docker-compose.voice.yml):
#   docker compose -f docker-compose.voice.yml up -d
# ============================================================

# ── Stage 1: Builder ────────────────────────────────────────
FROM node:20-slim AS builder

WORKDIR /app

COPY package.json package-lock.json ./
RUN npm ci

COPY tsconfig.json ./
COPY src/ ./src/

RUN npm run build

# ── Stage 2: Runtime ────────────────────────────────────────
FROM node:20-slim AS runtime

# System deps: Playwright/Chromium + Tesseract OCR + ffmpeg (for voice audio conversion)
RUN apt-get update && apt-get install -y --no-install-recommends \
    ca-certificates \
    curl \
    wget \
    ffmpeg \
    fonts-liberation \
    libatk-bridge2.0-0 \
    libatk1.0-0 \
    libcairo2 \
    libcups2 \
    libdbus-1-3 \
    libdrm2 \
    libexpat1 \
    libgbm1 \
    libglib2.0-0 \
    libgtk-3-0 \
    libnspr4 \
    libnss3 \
    libpango-1.0-0 \
    libpangocairo-1.0-0 \
    libx11-6 \
    libx11-xcb1 \
    libxcb1 \
    libxcomposite1 \
    libxdamage1 \
    libxext6 \
    libxfixes3 \
    libxrandr2 \
    libxrender1 \
    libxss1 \
    libxtst6 \
    xdg-utils \
    tesseract-ocr \
    && rm -rf /var/lib/apt/lists/*

WORKDIR /app

# Production deps only
COPY package.json package-lock.json ./
RUN npm ci --omit=dev

# Install Playwright browser binaries
RUN npx playwright install chromium --with-deps 2>/dev/null || true

# ── Voice: Piper TTS ────────────────────────────────────────
# Install Piper binary + shared libs + Korean voice model
RUN ARCH=$(dpkg --print-architecture) && \
    if [ "$ARCH" = "amd64" ]; then PIPER_ARCH="x86_64"; \
    elif [ "$ARCH" = "arm64" ]; then PIPER_ARCH="aarch64"; \
    elif [ "$ARCH" = "armhf" ]; then PIPER_ARCH="armv7l"; \
    else PIPER_ARCH="$ARCH"; fi && \
    cd /tmp && \
    curl -L "https://github.com/rhasspy/piper/releases/download/2023.11.14-2/piper_linux_${PIPER_ARCH}.tar.gz" -o piper.tar.gz && \
    tar xzf piper.tar.gz && \
    cp piper/piper /usr/local/bin/piper && \
    chmod +x /usr/local/bin/piper && \
    cp piper/libespeak-ng.so* /usr/local/lib/ && \
    cp piper/libonnxruntime.so* /usr/local/lib/ && \
    cp piper/libpiper_phonemize.so* /usr/local/lib/ && \
    cp piper/espeak-ng /usr/local/lib/ && \
    cp -r piper/espeak-ng-data /usr/local/lib/ && \
    ldconfig && \
    rm -rf piper piper.tar.gz && \
    mkdir -p /usr/local/share/piper-voices && \
    curl -L "https://huggingface.co/neurlang/piper-onnx-kss-korean/resolve/main/piper-kss-korean.onnx" \
    -o /usr/local/share/piper-voices/ko_KR-kss-medium.onnx && \
    curl -L "https://huggingface.co/neurlang/piper-onnx-kss-korean/raw/main/piper-kss-korean.onnx.json" \
    -o /usr/local/share/piper-voices/ko_KR-kss-medium.onnx.json

# ── Voice: whisper.cpp (STT) ────────────────────────────────
# Build from source (no pre-built Linux binary available)
RUN apt-get update && apt-get install -y --no-install-recommends cmake git build-essential && \
    cd /tmp && \
    git clone https://github.com/ggml-org/whisper.cpp.git && \
    cd whisper.cpp && \
    cmake -B build && \
    cmake --build build --config Release -j$(nproc) && \
    cp build/bin/whisper-cli /usr/local/bin/whisper-cli && \
    rm -rf /tmp/whisper.cpp && \
    apt-get remove -y git build-essential && \
    apt-get autoremove -y && \
    rm -rf /var/lib/apt/lists/* && \
    mkdir -p /usr/local/share/whisper-models && \
    curl -L "https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-base.bin" \
    -o /usr/local/share/whisper-models/ggml-base.bin

# ── Voice: espeak-ng-data symlink ────────────────────────────
# Piper expects espeak-ng-data at /usr/share/espeak-ng-data
RUN ln -sf /usr/local/lib/espeak-ng-data /usr/share/espeak-ng-data

# ── Voice: Telegram DNS fix ───────────────────────────────────
# Some networks block 149.154.166.110; use working IP instead
RUN echo "149.154.167.220 api.telegram.org" >> /etc/hosts

# ── App ──────────────────────────────────────────────────────
COPY --from=builder /app/dist ./dist

# Static web UI
COPY web-ui/ ./web-ui/

# Data directories (overridden by volumes in compose)
RUN mkdir -p /data/workspace /data/logs /root/.localclaw

# ── Environment defaults ──────────────────────────────────────
ENV NODE_ENV=production \
    DOCKER_CONTAINER=true \
    SMALLCLAW_DATA_DIR=/data \
    SMALLCLAW_WORKSPACE_DIR=/data/workspace \
    GATEWAY_PORT=18789 \
    GATEWAY_HOST=0.0.0.0 \
    PLAYWRIGHT_BROWSERS_PATH=/root/.cache/ms-playwright \
    \
    SMALLCLAW_PROVIDER=ollama \
    \
    OLLAMA_HOST=http://ollama:11434 \
    \
    LM_STUDIO_ENDPOINT=http://host.docker.internal:1234 \
    LM_STUDIO_API_KEY="" \
    LM_STUDIO_MODEL="" \
    \
    LLAMA_CPP_ENDPOINT=http://host.docker.internal:8080 \
    LLAMA_CPP_MODEL="" \
    \
    OPENAI_API_KEY="" \
    OPENAI_MODEL=gpt-4o \
    \
    CODEX_MODEL=gpt-5.3-codex \
    \
    PIPER_MODEL_PATH=/usr/local/share/piper-voices/ko_KR-kss-medium.onnx \
    PIPER_CONFIG_PATH=/usr/local/share/piper-voices/ko_KR-kss-medium.onnx.json \
    WHISPER_MODEL_PATH=/usr/local/share/whisper-models/ggml-base.bin

EXPOSE 18789

HEALTHCHECK --interval=30s --timeout=10s --start-period=30s --retries=3 \
    CMD curl -f http://localhost:18789/health || exit 1

CMD ["node", "dist/cli/index.js", "gateway"]