Dockerfile

FROM python:3.10-bullseye
RUN --mount=type=cache,target=/var/cache/apt,sharing=locked \
    apt-get update --fix-missing  && apt-get upgrade -y && \
    apt-get install -y --fix-missing --no-install-recommends git build-essential gcc g++ portaudio19-dev ffmpeg libportaudio2 libasound-dev python3 python3-pip gcc wget curl && \
    apt-get clean && rm -rf /var/lib/apt/lists/* && \
    python3 -m pip install --upgrade pip --no-cache-dir
WORKDIR /app

# Install PyTorch CPU version
RUN pip install torch==2.6.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cpu --no-cache-dir

# Install numpy and Cython for pkuseg (required by chatterbox-tts)
RUN pip install numpy==1.25.2 Cython "setuptools>=78.1.1" --no-cache-dir
# Install pkuseg separately (required by chatterbox-tts)
RUN pip install pkuseg==0.0.25 --no-build-isolation --no-cache-dir

COPY requirements.txt .
RUN --mount=type=cache,target=/root/.cache/pip,sharing=locked \
    python3 -m pip install --no-cache-dir -r requirements.txt

# Install chatterbox-tts with --no-deps to bypass transformers==4.46.3 pin
# This allows us to use transformers>=4.53.0 for security fixes
RUN pip install chatterbox-tts --no-deps --no-cache-dir

# Install xllamacpp CPU version
RUN pip install xllamacpp==0.2.12 --force-reinstall --no-cache-dir

COPY . .
ENV HOST=0.0.0.0 \
    TOKENIZERS_PARALLELISM=false
EXPOSE 8091
# Use start.py which runs precache once, then starts uvicorn workers
CMD ["python", "start.py"]