TypeError

[xiaojian10]

I tried to run the code you shared, but when I type the address of my voice file, the code will go wrong. Just like


Is this because the format of my input file is wrong? Or what are the specific requirements for the format of the input voice file? For example, its sample rate, bit rate, int16 or float64...I hope that answer my question in your free time , I would be grateful.

[zhr1201]

It is basically saying that your version of np cannot take float64 and automatically cast that into a integer. So probably you want manually do that by using type casting.

[xiaojian10]

Thank you very much for your suggestion, I have solved this problem.

[xiaojian10]

I have been reading your articles and code. I found that the three audio inputs in the code are not mentioned in your article. I think forward beamformer output audio and backward beamformer output audio correspond to the output of Beam1 and Beam2 in the article, and reference beamformer output audio corresponds to the reference signals for mics in the article. Is this understanding correct?

[zhr1201]

Yeah, that's right!

[xiaojian10]

When I run SENN_train.py，It has the following error：


I tried a lot of ways to solve this problem, but all failed. May I ask how to solve this problem, can you give me some advice?

[zhr1201]

Sorry for the delay. I have encountered this type of error before. It is because of tf version compatibility. So please change that to the api of your current tf version.

[xiaojian10]

I really appreciate your help. I believe that with your help, I can achieve it very quickly.