LEE YOON HYUNG

Considering that the query and key have a timestep of 1 from the perspective of 1d convolution, I think that it is totally same with using the fully connected layer....

Hi. Thank you for your implementation, and I have a question about the optimizer. It seems that you use Adam optimizer with lr=1e-3 and amsgrad=True. Why you choose the options...

1. epsilon을 std와 distrib 둘다 더해주었습니다. 2. model.py에서 t, f방향으로 나눠주는 방법이 제가 이해한 바랑 살짝 다른 거 같아요. self.f_div = f_div[hp.n_tiers], self.t_div = t_div[hp.n_tiers]를 self.f_div = f_div[hp.n_tiers+1], self.t_div =...

bucket_by_sequence_length는 제 생각대로 구현은 해봤는데 wav 파일들을 전부 로드해서 한번 길이를 구하는 과정도 너무 오래 걸리네요. 수정이 필요해 보입니다.

FastSpeech2 trained using LibriTTS dataset

3

Hi, my name is Yoonhyung Lee, who is studying Text-to-Speech. Thank you for your nice implementation of FastSpeech2. It helped me a lot to study it, but a question occurred...

relative_multi_head_attention.py 구현 수정 제안

1

# ❓ Questions & Help relative_multi_head_attention.py 구현 수정 제안드립니다. ## Details 1. self.sqrt_dim: math.sqrt(dim) -> math.sqrt(d_head) 2. 현재의 _relative_shift는 relative position difference가 -1인 경우 T만큼 떨어져있다고 계산이 되고 있는 것...

Why stop_gradient operator is used in diffstride?

1

Firstly, I want to appreciate for sharing great work with other researchers. I write this issue because I have a question about using the stop_gradient operator. When I saw other...

Thank you for your excellent work on this project! I would like to inquire about the dataset used for training the CRF-based G2P model. Could you please provide details on...