fix: Dockerfile

Dockerfile

@@ -1,6 +1,6 @@
 FROM python:3.12-slim
 
-RUN apt-get update && apt-get install -y build-essential ffmpeg  && rm -rf /var/lib/apt/lists/*
+RUN apt-get update && apt-get install -y build-essential ffmpeg libsndfile1 && rm -rf /var/lib/apt/lists/*
 
 COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/
 
@@ -9,7 +9,20 @@ COPY . .
 RUN uv sync
 RUN uv sync --extra vllm
 
+ENV PYTHONUNBUFFERED=1
+ENV HF_HOME="/app/.cache/huggingface"
 ENV ASR_MODEL_PATH="Qwen/Qwen3-ASR-1.7B"
+ENV GPU_MEMORY_UTILIZATION=0.8
+ENV MAX_NEW_TOKENS=32
+ENV UNFIXED_CHUNK_NUM=4
+ENV UNFIXED_TOKEN_NUM=5
+ENV CHUNK_SIZE_SEC=1.0
 
 EXPOSE 5000
-CMD ["sh", "-c", "uv run examples/api_unified_fastapi.py --asr-model-path $ASR_MODEL_PATH"]
+CMD ["sh", "-c", "uv run examples/api_unified_fastapi.py \
+    --asr-model-path $ASR_MODEL_PATH \
+    --gpu-memory-utilization $GPU_MEMORY_UTILIZATION \
+    --max-new-tokens $MAX_NEW_TOKENS \
+    --unfixed-chunk-num $UNFIXED_CHUNK_NUM \
+    --unfixed-token-num $UNFIXED_TOKEN_NUM \
+    --chunk-size-sec $CHUNK_SIZE_SEC"]

README_BW.md

@@ -23,8 +23,9 @@ docker pull harbor.bwgdi.com/library/qwen3-asr:0.0.1
 
 # Run with custom model path
 # -e ASR_MODEL_PATH: Model name or local path inside container
-docker run -d --restart always -p 5051:5000 --gpus all \
+docker run -d --restart always -p 5051:5000 --gpus '"device=2"' \
   -e ASR_MODEL_PATH="Qwen/Qwen3-ASR-1.7B" \
+  -e GPU_MEMORY_UTILIZATION=0.8 \
   --mount type=bind,source=/path/to/your/models,target=/models \
   harbor.bwgdi.com/library/qwen3-asr:0.0.1
 ```