Dockerfile

FROM python:3.10-slim

RUN apt-get update && apt-get install -y --no-install-recommends ffmpeg && rm -rf /var/lib/apt/lists/*

WORKDIR /code

ENV PYTHONUNBUFFERED=1

# Install PyTorch CPU-only first (from PyTorch mirror, avoids huge CUDA download)
RUN pip install --no-cache-dir --retries 10 --timeout 120 \
    torch==2.2.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cpu

COPY requirements.txt .
# Install remaining deps (torch/torchaudio already satisfied, won't re-download)
RUN pip install --no-cache-dir --retries 10 --timeout 120 -r requirements.txt

COPY . .

RUN chmod -R 777 /code

# Pre-download Wav2Vec2 model during build so startup doesn't need network
RUN python -c "from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor; \
    Wav2Vec2Processor.from_pretrained('imvladikon/wav2vec2-large-xlsr-53-hebrew'); \
    Wav2Vec2ForCTC.from_pretrained('imvladikon/wav2vec2-large-xlsr-53-hebrew')"

EXPOSE 7860

CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]