search

zzw922cn / Automatic_Speech_Recognition

End-to-end Automatic Speech Recognition for Madarian and English in Tensorflow
MIT License
2.84k stars 538 forks source link

what should I do when I have run the program successfuly #44

Open Ostnie opened 6 years ago

Ostnie commented 6 years ago

I don't know how to draw the PER form like the author, and I also want to use the trained model to identify my own voice data, and I don't know how to do it，many thx！

zzw922cn commented 6 years ago

@Ostnie You can set the mode=test to evaluate your trained model.

Ostnie commented 6 years ago

@zzw922cn 1.My program has saved a lot of model in data/log/timit/phn/save ,and which model will the program use?and if I want to use my own data,do I need to preprocess ? 2.I don't know how to draw the PER form like the you ,I want to know how the picture in your file is produced. many thanks,and I have written two letters to your mailbox, and I want to know whether you have received them,one in Chinese and one in English.

zzw922cn commented 6 years ago

@Ostnie you can change the content of checkpoint file, the first line should be what model you want to use..

About PER, it seems in my code, i didn't show how to draw it. But you can keep record the PER and then plot.

Ostnie commented 6 years ago

@zzw922cn sorry，I seem to have forgotten to say that I am running a timit program，maybe can help you better solve the problem。I have three questions and I hope you can help me. 1. I tried to change the last digit of the file in the first line of the checkpoint, but it looks like no matter which one of my last test modes I chose to pick, the result is around 0.75, but I've dropped to 0.4 during the 100 iterations of training, and I do not know why Will be like this。

2.Do I still want to know that testing only one voice data? The result of data processing told me there is a lot of test data, I think the procedure has also been loaded during the operation of all the test data, but I only saw in the result of a speech recognition results,and I don't konw why? 3.When the program is running with GPU,each time the number of changes in the batch will output such a sentence： 2017-11-03 13:11:11.592943：I tensorflow/core/commom_runtime/gpu/gpu_device.cc:977]Creating temsorflow device (/gpu:0)->(device:0,name:GeForce GTX 1080 Ti,pci bus id : 0000:03:00.0 This sentence repeated hundreds of times, I am very distressed about this, you can cancel it, too affect me to see the operation of the process