Bug Report: UVR Model Re-downloads on Each Run in Docker v1.0.7; torchaudioAttributeError in latest

[merlinmxn]

Using the Docker v1.0.7-fef3d69 image, every time I perform background music separation now, it re-downloads the UVR-MDX-NET-Inst_HQ_4 model. I have already set up persistence, but each time I execute the separation of background music and vocals, it downloads the model again. Docker doesn't even need to be restarted; every time a song is submitted, it re-downloads the model.
As for the latest version, it throws this error: AttributeError: module 'torchaudio' has no attribute 'AudioMetaData' #606, so I can't test it. My host machine is: NVIDIA-SMI 560.28.03, Driver Version: 560.28.03, CUDA Version: 12.6.

[merlinmxn]

services:
whisper-webui:
container_name: whisper-webui
build: .
image: jhj0517/whisper-webui:v1.0.7-fef3d69
volumes:
- ./models:/Whisper-WebUI/models
- ./outputs:/Whisper-WebUI/outputs
- ./configs:/Whisper-WebUI/configs
ports:
- "7860:7860"
stdin_open: true
tty: true
entrypoint: ["python", "app.py", "--server_port", "7860", "--server_name", "0.0.0.0"]
runtime: nvidia
environment:
- LANG=zh_CN.UTF-8
- TZ=Asia/Shanghai
- NVIDIA_VISIBLE_DEVICES=0 # 指定使用第0块GPU
- NVIDIA_DRIVER_CAPABILITIES=all
privileged: true
restart: no

[quinn0823]

I think #577 has already raised this issue and pointed out the cause of the error. Refer to #577 (comment) for a temporary solution.

[merlinmxn]

I think #577 has already raised this issue and pointed out the cause of the error. Refer to #577 (comment) for a temporary solution.

Thank you. It's settled

services:
whisper-webui:
container_name: whisper-webui
build: .
image: jhj0517/whisper-webui:v1.0.7-fef3d69
volumes:
- ./models:/Whisper-WebUI/models
- ./outputs:/Whisper-WebUI/outputs
- ./configs:/Whisper-WebUI/configs
- ./models/UVR/MDX_Net_Models/UVR-MDX-NET-Inst_3.onnx:/Whisper-WebUI/venv/lib/python3.11/site-packages/uvr/models_dir/mdx/weights/UVR-MDX-NET-Inst_3/UVR-MDX-NET-Inst_3.onnx
- ./models/UVR/MDX_Net_Models/UVR-MDX-NET-Inst_HQ_4.onnx:/Whisper-WebUI/venv/lib/python3.11/site-packages/uvr/models_dir/mdx/weights/UVR-MDX-NET-Inst_HQ_4/UVR-MDX-NET-Inst_HQ_4.onnx
ports:
- "7860:7860"
stdin_open: true
tty: true
entrypoint: ["python", "app.py", "--server_port", "7860", "--server_name", "0.0.0.0"]
runtime: nvidia
environment:
- LANG=zh_CN.UTF-8
- TZ=Asia/Shanghai
- NVIDIA_VISIBLE_DEVICES=0
- NVIDIA_DRIVER_CAPABILITIES=all
# privileged: true
restart: always