Juntae, Kim

Sorry for absent of detail description of specification of training dataset, But, it is very simple. You can find the dataformat by investigating the data in /data/raw/train or /data/raw/valid The...

You don't have to conduct framing on the label. The needed label is just sample based label. For example if speech signal has 10,000 samples. The label also should have...

Your guess is correct, the 1 corresponds to speech and 0 corresponds to the non-speech the plot is like as below:  Note that if the speech data has noise,...

Sorry for the late answer because I'm too busy in these days. The separation means that 'not sharing gradients'. In this code, you can see that gradients of reinforce part...

Thank you! please look my code in your spare time and it will be very good to discuss about it.

@JasonZhao001 @GodOfProbability Hi Jason, I read your comment very interestingly, I also solve the stop_gradient problem, but cannot achieve high performance as like you (in my case, about 94%, translated...

@JasonZhao001 Thank your for your kind comment, I also agree with you. However, in this code, the baseline is implemented like below, baseline = tf.sigmoid(tf.matmul(hiddensState, Wb_h_b)+Bb_h_b) and, both Wb_h_b and...

@JasonZhao001 Thank you! I missed that part. I'll expect your implementation have a good day!

Also, your weighted_relative_edit_error includes the information about ground truth, not n-best only.