How do I bind the PESQ function to the "Denoise Speech Using Deep Learning Networks" instance?

I ran this example but now I want to measure the sound improvement with the PESQ function. But when I run the function, it throws an error. I will be glad if you tell me where I am going wrong.

https://www.mathworks.com/help/deeplearning/ug/denoise-speech-using-deep-learning-networks.html#DenoiseSpeechUsingDeepLearningNetworksExample-7

ref_wav=cleanAudio;
deg_wav=double(denoisedAudioFullyConnected);

[pesq_mos]= pesq(ref_wav, deg_wav)

Error using audioread (line 105)
Expected input to be one of these types:

char, string

Instead its type was double.

Error in pesq (line 42)
[ref_data,sampling_rate]= audioread(ref_wav);

Answer 1

[yanqi liu]

yes，sir，if use single vector as input，may be not need read，just use them as ref_data、deg_data