LEE YOON HYUNG comments

Results 7 comments of


                                            LEE YOON HYUNG

Did you try to replace LSA in Tacotron2(Nvidia) with this DCA implemention ?

Hi. I had similar problems when I tried to apply DCA to NVIDIA/tacotron2. (e.g. it can not learn the alignment, output NaN loss) Then, when I use beta=6.3 and grad_thresgold=0.05...

Did you try to replace LSA in Tacotron2(Nvidia) with this DCA implemention ?

(1) Higher beta rather decelerates the movement of alignment at each decoder step. (average αn/(α + β) step) (2) When I apply noam_lr_scheduling, it seems really helpful for the training.

FastSpeech2 trained using LibriTTS dataset

I think it is quite natural learning multi-speaker TTS is more difficult. Thank you for your reply :D

train.py fails with error "too many indices for tensor of dimension 2"

Do you change the file 'functional.py' in original pytorch package following the description at the bottom of the README.md? I think it can cause the problem. Also, I revise this...

Better explanation of "I failed to train the fastspeech"

According to the writers of the fastspeech, it is important to use proper alignments in the training. When I implemented transformer-tts at first, I failed to implement it perfectly, and...

Great to see the first imp of AlignTTS in github!

Thanks for trying to cheer us up, but we can not still succeed to train the AlignTTS. We are having a hard time training the MDN because of the numerical...

[Help]: more duration predictor details & '50' meaning of target len in text2semantic of MaskGCT

I believe, though I'm not entirely certain, that the 50 refers to the sampling rate of semantic tokens. (50 tok/sec)