Dockerfile

FROM nvidia/cuda:12.4.1-devel-ubuntu22.04

ENV DEBIAN_FRONTEND=noninteractive
ENV PYTHONUNBUFFERED=1
ENV HF_HOME=/app/model_cache
ENV TORCH_HOME=/app/torch_cache

# Install Python 3.12 + system deps
RUN apt-get update && apt-get install -y \
    software-properties-common \
    git \
    curl \
    ffmpeg \
    libsndfile1 \
    && add-apt-repository ppa:deadsnakes/ppa \
    && apt-get update \
    && apt-get install -y python3.12 python3.12-venv python3.12-dev \
    && curl -sS https://bootstrap.pypa.io/get-pip.py | python3.12 \
    && ln -sf /usr/bin/python3.12 /usr/bin/python \
    && ln -sf /usr/bin/python3.12 /usr/bin/python3 \
    && rm -rf /var/lib/apt/lists/*

# Remove conflicting system packages
RUN apt-get update && apt-get remove -y python3-blinker || true \
    && rm -rf /usr/lib/python3/dist-packages/blinker* \
    && rm -rf /var/lib/apt/lists/*

# Install Python dependencies
COPY requirements.txt /app/requirements.txt
WORKDIR /app
RUN pip install --no-cache-dir --break-system-packages -r requirements.txt

# Install flash-attn for faster inference (optional, build may fail on some GPUs)
RUN MAX_JOBS=4 pip install flash-attn --no-build-isolation \
    || echo "flash-attn install failed, continuing without it"

# Pre-download ASR model into the image (no persistent storage available on HF Spaces)
RUN mkdir -p /app/model_cache /app/torch_cache && \
    python -c "\
from huggingface_hub import snapshot_download; \
snapshot_download('Qwen/Qwen3-ASR-1.7B', cache_dir='/app/model_cache')" \
    && echo "ASR model pre-downloaded"

# Pre-download Silero VAD model
RUN python -c "\
import torch; \
torch.hub.load('snakers4/silero-vad', 'silero_vad', trust_repo=True); \
print('Silero VAD cached')" || echo "VAD pre-download failed, will retry at runtime"

# Copy application code
COPY . /app

EXPOSE 7860

# Health check — long start period for model download + load
HEALTHCHECK --interval=30s --timeout=30s --start-period=600s --retries=3 \
    CMD curl -f http://localhost:7860/health || exit 1

CMD ["python", "app.py"]
# Rebuild trigger: Wed Apr 22 18:13:56 PKT 2026