Issue running sound classification demo about open_model_zoo HOT 9 CLOSED

Why do you use that much / in paths: /home///..., /home////...? That's probably the reason for the error

from open_model_zoo.

those were my machine credentials, i deleted them, since I don't wan them out in public :-)
paths are all correct

from open_model_zoo.

I see only two possible reasons:

1. There's version mismatch between mo and OpenVINO Runtime. Run mo --version and compare that to what the demo prints.
2. File path is incorrect. You can try running python3 -c "import openvino.runtime as ov; ov.Core().read_model('XML_PATH')" to make sure there's nothing in demo itself causing the problem.
   BTW, the usual path is public/aclnet/FP16/aclnet.xml maybe you miss FP16.

from open_model_zoo.

Thank you. It was version mismatch, but now I am getting a different error. Any pointers ?
[ INFO ] OpenVINO Runtime
[ INFO ] build: 2023.0.1-11005-fa1c41994f3-releases/2023/0
[ INFO ] Reading model /home////public/aclnet/aclnet_des_53.xml
Traceback (most recent call last):
File "sound_classification_demo.py", line 193, in
main()
File "sound_classification_demo.py", line 151, in main
batch_size, channels, one, length = model.inputs[0].shape
ValueError: get_shape was called on a descriptor::Tensor with dynamic shape

from open_model_zoo.

Both mo & openvino runtime have the same versions. I generated the IR by running the mo command - mo --input_model aclnet_des_53.onnx

[ INFO ] OpenVINO Runtime
[ INFO ] build: 2023.0.1-11005-fa1c41994f3-releases/2023/0

mo --version
Version of Model Optimizer is: 2023.0.1-11005-fa1c41994f3-releases/2023/0

from open_model_zoo.

The demo supports static shapes only.

One option is to update the demo to support dynamic shapes. But I'm not going to help here.

Another option is to set shapes to be static at mo stage. Regenerate IR by running mo --input_model aclnet_des_53.onnx --input_shape=[1,1,1,16000]. Assuming your aclnet_des_53.onnx is similar to one we have in https://github.com/openvinotoolkit/open_model_zoo/blob/f08b9b080b240b4abd08830542ef532deeaf1da3/models/public/aclnet/model.yml#L33C2-L33C2

from open_model_zoo.

I now get a channel issue. Any suggestions on where I can get the sample audio from openvino repo? I got this input audio from google.

[ INFO ] 	build: 2023.0.1-11005-fa1c41994f3-releases/2023/0
[ INFO ] Reading model /home////public/aclnet/aclnet_des_53.xml
[ INFO ] The model /home////public/aclnet/aclnet_des_53.xml is loaded to CPU
Traceback (most recent call last):
  File "sound_classification_demo.py", line 193, in <module>
    main()
  File "sound_classification_demo.py", line 171, in main
    audio = AudioSource(args.input, channels=channels, samplerate=args.sample_rate)
  File "sound_classification_demo.py", line 67, in __init__
    raise RuntimeError("Audio has unsupported number of channels - {} (expected {})"
RuntimeError: Audio has unsupported number of channels - 2 (expected 1)

from open_model_zoo.

https://storage.openvinotoolkit.org/models_contrib/speech/2021.2/librispeech_s5/how_are_you_doing_today.wav

from open_model_zoo.

Thank a lot @Wovchena. Demo worked :-)

from open_model_zoo.