Add: VibeVoice ASR セットアップスクリプト一式

2026-02-24 01:21:33 +00:00
parent 2d753f114f
commit 1fb76254e9
15 changed files with 4531 additions and 0 deletions
@@ -0,0 +1,61 @@
+# VibeVoice-ASR for DGX Spark (ARM64, Blackwell GB10, sm_121)
+# Based on NVIDIA PyTorch container for CUDA 13.1 compatibility
+
+ARG TARGETARCH
+FROM nvcr.io/nvidia/pytorch:25.11-py3 AS base
+
+LABEL maintainer="VibeVoice-ASR DGX Spark Setup"
+LABEL description="VibeVoice-ASR optimized for DGX Spark (ARM64, CUDA 13.1)"
+
+# Set environment variables
+ENV DEBIAN_FRONTEND=noninteractive
+ENV PYTHONDONTWRITEBYTECODE=1
+ENV PYTHONUNBUFFERED=1
+
+# PyTorch CUDA settings for DGX Spark
+ENV PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True
+ENV USE_LIBUV=0
+
+# Set working directory
+WORKDIR /workspace
+
+# Install system dependencies including FFmpeg for demo
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    ffmpeg \
+    git \
+    curl \
+    && apt-get clean \
+    && rm -rf /var/lib/apt/lists/*
+
+# Install flash-attn if not already present
+RUN pip install --no-cache-dir flash-attn --no-build-isolation || true
+
+# Clone and install VibeVoice
+RUN git clone https://github.com/microsoft/VibeVoice.git /workspace/VibeVoice && \
+    cd /workspace/VibeVoice && \
+    pip install --no-cache-dir -e .
+
+# Create test script and patched demo with MKV support
+COPY test_vibevoice.py /workspace/test_vibevoice.py
+COPY vibevoice_asr_gradio_demo_patched.py /workspace/VibeVoice/demo/vibevoice_asr_gradio_demo.py
+
+# Install real-time ASR dependencies
+COPY requirements-realtime.txt /workspace/requirements-realtime.txt
+RUN pip install --no-cache-dir -r /workspace/requirements-realtime.txt
+
+# Copy real-time ASR module and startup scripts
+COPY realtime/ /workspace/VibeVoice/realtime/
+COPY static/ /workspace/VibeVoice/static/
+COPY run_all.sh /workspace/VibeVoice/run_all.sh
+COPY run_realtime.sh /workspace/VibeVoice/run_realtime.sh
+RUN chmod +x /workspace/VibeVoice/run_all.sh /workspace/VibeVoice/run_realtime.sh
+
+# Set default working directory to VibeVoice
+WORKDIR /workspace/VibeVoice
+
+# Expose Gradio port and WebSocket port
+EXPOSE 7860
+EXPOSE 8000
+
+# Default command: Launch Gradio demo with MKV support
+CMD ["python", "demo/vibevoice_asr_gradio_demo.py", "--model_path", "microsoft/VibeVoice-ASR", "--host", "0.0.0.0"]