training fastspeech2 on persian dataset

[arefekh]

Hi! We trained fastspeech2 network with Persian dataset. The point that caught our attention after the training of the network was that if we add english g2p outputs to the persian valid symbols, the quality of the output audio will increase. This is while we checked the dump/train/ids files and found that none of the english g2p outputs id are in these files. Do you have an opinion in this regard that can help us in choosing the right collection of valid symbols?