求助问题：ValueError: max() arg is an empty sequence

[Lock1994]

分割音频报ValueError: max() arg is an empty sequence错，是不是文件太大了？
在do_emb_extraction这步，抽取的音频长度都是0，更换了几个wav文件问题都一样
Rank 0 processing: 0%| | 0/1 [00:00<?, ?it/s]D:\AIGC\3D-Speaker\speakerlab\bin\1\3.wav
tensor([[0., 0., 0., ..., 0., 0., 0.]])
[]
0
Rank 0 processing: 0%| | 0/1 [00:03<?, ?it/s]
Traceback (most recent call last):
File "D:\AIGC\3D-Speaker\speakerlab\bin\infer_diarization.py", line 490, in
main()
File "D:\AIGC\3D-Speaker\speakerlab\bin\infer_diarization.py", line 487, in main
mp.spawn(main_process, nprocs=args.nprocs, args=(args.nprocs, args, wav_list))
File "C:\Users\aaa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\torch\multiprocessing\spawn.py", line 282, in spawn
return start_processes(fn, args, nprocs, join, daemon, start_method="spawn")
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\aaa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\torch\multiprocessing\spawn.py", line 238, in start_processes
while not context.join():
^^^^^^^^^^^^^^
File "C:\Users\aaa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\torch\multiprocessing\spawn.py", line 189, in join
raise ProcessRaisedException(msg, error_index, failed_process.pid)
torch.multiprocessing.spawn.ProcessRaisedException:

-- Process 0 terminated with the following error:
Traceback (most recent call last):
File "C:\Users\aaa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\torch\multiprocessing\spawn.py", line 76, in _wrap
fn(i, *args)
File "D:\AIGC\3D-Speaker\speakerlab\bin\infer_diarization.py", line 437, in main_process
ouput = diarization(wav_path)
^^^^^^^^^^^^^^^^^^^^^
File "D:\AIGC\3D-Speaker\speakerlab\bin\infer_diarization.py", line 202, in call
embeddings = self.do_emb_extraction(chunks, wav_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\AIGC\3D-Speaker\speakerlab\bin\infer_diarization.py", line 253, in do_emb_extraction
wavs = torch.stack(wavs).unsqueeze(1)
^^^^^^^^^^^^^^^^^
RuntimeError: stack expects a non-empty TensorList

[wanghuii1]

这些文件是静音吗，如果vad没有检测到有效音频会出现这个错误

[Lock1994]

这些文件是静音吗，如果vad没有检测到有效音频会出现这个错误
不是静音文件，是有人声的，我试过3个wav文件，都在Vad这里过不去
File "C:\Users\aaa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\funasr\models\fsmn_vad_streaming\model.py", line 522, in GetFrameState
speech_prob = math.log(sum_score)
^^^^^^^^^^^^^^^^^^^
ValueError: math domain error
有测试样例的WAV文件吗？我再验证下

[wanghuii1]

你运行3D-Speaker/egs/3dspeaker/speaker-diarization/run_audio.sh，这个会下载测试音频

[wanghuii1]

这些文件是静音吗，如果vad没有检测到有效音频会出现这个错误
不是静音文件，是有人声的，我试过3个wav文件，都在Vad这里过不去
File "C:\Users\aaa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\funasr\models\fsmn_vad_streaming\model.py", line 522, in GetFrameState
speech_prob = math.log(sum_score)
^^^^^^^^^^^^^^^^^^^
ValueError: math domain error
有测试样例的WAV文件吗？我再验证下

看错误，是vad调用出错了，没有出结果，你可以单独调用这个模型看看https://modelscope.cn/models/iic/speech_fsmn_vad_zh-cn-16k-common-pytorch

[Lock1994]

这些文件是静音吗，如果vad没有检测到有效音频会出现这个错误
不是静音文件，是有人声的，我试过3个wav文件，都在Vad这里过不去
File "C:\Users\aaa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\funasr\models\fsmn_vad_streaming\model.py", line 522, in GetFrameState
speech_prob = math.log(sum_score)
^^^^^^^^^^^^^^^^^^^
ValueError: math domain error
有测试样例的WAV文件吗？我再验证下

看错误，是vad调用出错了，没有出结果，你可以单独调用这个模型看看https://modelscope.cn/models/iic/speech_fsmn_vad_zh-cn-16k-common-pytorch

更换了VAD代码，新的问题
File "D:\AIGC\3D-Speaker\speakerlab\bin\infer_diarization.py", line 256, in do_emb_extraction
wavs = torch.stack(wavs).unsqueeze(1)
^^^^^^^^^^^^^^^^^
RuntimeError: stack expects a non-empty TensorList

[wanghuii1]

是指VAD代码更换你自己的吗

[Lock1994]

是指VAD代码更换你自己的吗
是全部更换为https://modelscope.cn/models/iic/speech_fsmn_vad_zh-cn-16k-common-pytorch/summary的VAD，问题如下，依然输不出结果
funasr version: 1.2.2.
2024-12-31 18:12:47,906 - modelscope - WARNING - No preprocessor field found in cfg.
2024-12-31 18:12:47,906 - modelscope - WARNING - No val key and type key found in preprocessor domain of configuration.json file.
2024-12-31 18:12:47,906 - modelscope - WARNING - Cannot find available config to build preprocessor at mode inference, current config: {'model_dir': 'iic/speech_fsmn_vad_zh-cn-16k-common-pytorch'}. trying to build by task and model information.
2024-12-31 18:12:47,906 - modelscope - WARNING - No preprocessor key ('funasr', 'voice-activity-detection') found in PREPROCESSOR_MAP, skip building preprocessor.
2024-12-31 18:12:47,913 - modelscope - INFO - cuda is not available, using cpu instead.
rtf_avg: 0.082: 100%|██████████| 1/1 [00:00<00:00, 1.17it/s]
[{'key': '4', 'value': []}]

[wanghuii1]

infer_diarization.py使用的vad就是https://modelscope.cn/models/iic/speech_fsmn_vad_zh-cn-16k-common-pytorch，该vad的输出结果是空[{'key': '4', 'value': []}]，说明该vad模型没有检测出有效语音，可以使用其他能检测出你的音频中语音的vad模型进行替换。