Erdene-Ochir Tuguldur

Add `%tensorflow_version 1.x` at the top of the notebook and execute it. Colab now uses per default TF2, if you use this magic command, it will use TF1.

@jayaneetha do you have pretrained weights? or sample audios?

@nmfisher I have reimplemented the encoder outputting mel spectrograms and trained from scratch 150k steps on a custom dataset. It sounds OK with the universal Waveglow vocoder. It should be...

Currently download from Google Drive doesn't work anymore, as long `gdown` doesn't fix it, we can't do anything, see: https://github.com/wkentaro/gdown/issues/162

@thetroll288 i have started to fix the notebooks, the reason why the most notebooks didn't work was, different pytorch and `gdown` issues. Hope i will fix in the next few...

@SutirthaChakraborty you have to enable the BUILD_Python cmake configuration.

300 hours dataset is not Mongolian but Kalmyk (Western Mongolian language).

I have also a 5h dataset. I have tried to train with r=1 and batch_size=48 on 2 GPUs, but haven't got any alignment after 100K steps. With r=2, good alignment...

see also: https://github.com/Rayhane-mamah/Tacotron-2/issues/175 I will resume the training until it reaches 200K and report the result here :)

@begeekmyfriend have you got any alignment with `r=1`?