No clear speech

[syang1993]

No clear speech

[ErnstTmp]

Sure, the model is at https://www.dropbox.com/s/fl1vqfz6611s8zw/checkpoint234000.tar?dl=0

[ishandutta2007]

Can you share the model till now.

[MengLiPKU]

Do you have any idea now? I'm facing the same problem

[syang1993]

Hi, sorry for the late reply. In my early experiments, I found only when I use some unseen reference audio, the generated sometimes would become bad.

And I think the batch size will affect the quality and stability. I also find multi-gpu training for tacotron is better than single GPU model.

I guess you can reduce the head number (better token diversity) and modify the attention mechanism as we discussed before.

[ErnstTmp]

Thanks a lot for your help - batch size seem to be the problem. I had to reduce the batch size due to memory limits - and this seem to have caused the problems.

• I will retry with batchsize >= 32 and multi-gpu if possible.

[ZohaibAhmed]

@ErnstTmp Did you attempt it with multi-gpu at a higher batch size? Was keen on seeing if the results were better.

[ErnstTmp]

I switched temporarily to Tacotron-2, and that works with single GPU with more memory (16 GB) and batchsize 32.
Also, I could increasing the number of parallel outputs to be able to run batchsize 32 on a 12 GB GPU.