Dockerfile

FROM pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime

# Set environment variables
ENV PYTHONUNBUFFERED=1 \
    PYTHONDONTWRITEBYTECODE=1 \
    PIP_NO_CACHE_DIR=1 \
    PIP_DISABLE_PIP_VERSION_CHECK=1 \
    HF_HOME=/app/hf_cache

# Install system dependencies
RUN apt-get update && apt-get install -y --no-install-recommends \
    ffmpeg \
    sox \
    libsox-dev \
    git \
    curl \
    && rm -rf /var/lib/apt/lists/*

# Set working directory
WORKDIR /app

# Copy requirements first for better caching
COPY requirements.txt .

# Install Python dependencies
RUN pip install --no-cache-dir -r requirements.txt
RUN pip install https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.5.4/flash_attn-2.8.3+cu124torch2.5-cp311-cp311-linux_x86_64.whl


# Pre-download model (accepts shorthand like "0.6B" or full path)
ARG MODEL_NAME=0.6B
ENV MODEL_NAME=${MODEL_NAME}

# Copy config first to use MODEL_MAP for resolving shorthand
COPY app/config.py .

# Create cache directory and download model during build
RUN mkdir -p /app/hf_cache && \
    python -c "from config import get_model_name; from huggingface_hub import snapshot_download; snapshot_download(get_model_name())"

# Copy application code
COPY app/ .

# Create directories
RUN mkdir -p /app/voices /app/cache

# Set default environment variables
ENV HOST=0.0.0.0 \
    PORT=8880 \
    DEVICE=cuda:0 \
    VOICES_DIR=/app/voices \
    CACHE_DIR=/app/cache

# Expose port
EXPOSE 8880

# Health check
HEALTHCHECK --interval=30s --timeout=30s --start-period=120s --retries=3 \
    CMD curl -f http://localhost:8880/health || exit 1

# Run server
CMD ["python", "server.py"]