分离用fox的那个模型出错

(most recent call last): File "/workspace/rvc/infer/modules/uvr5/modules.py", line 72, in uvr pre_fun.path_audio( File "/workspace/rvc/infer/modules/uvr5/mdxnet.py", line 240, in path_audio self.pred.prediction(input, vocal_root, others_root, format) File "/workspace/rvc/infer/modules/uvr5/mdxnet.py", line 209, in prediction save_audio( File "/workspace/rvc/infer/lib/audio.py", line 47, in save_audio buf = float_np_array_to_wav_buf(audio, sr, f32) File "/workspace/rvc/infer/lib/audio.py", line 35, in float_np_array_to_wav_buf wavfile.write(buf, sr, wav.astype(np.float32)) File "/root/miniconda3/lib/python3.10/site-packages/scipy/io/wavfile.py", line 824, in write fmt_chunk_data = struct.pack('<HHIIHH', format_tag, channels, fs, TypeError: only integer scalar arrays can be converted to a scalar index

Start with the failing path in infer/modules/uvr5/modules.py, then follow path_audio and prediction in infer/modules/uvr5/mdxnet.py into save_audio and float_np_array_to_wav_buf in infer/lib/audio.py. Reproduce separation with the Fox model and inspect the values passed to wavfile.write. Done means the model completes separation and writes the vocal and other audio outputs without the reported TypeError.