size mismatch for encoder.embedding.weight exception when using miven model and NO Chinse voice but only noise

1)Use miven's model:
https://pan.baidu.com/s/1PI-hM3sn5wbeChRryX-RCQ code: 2021

2)Got exception
size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([70, 512]) from checkpoint, the shape in current model is torch.Size([75, 512]).

3)NOT ABLE to have Chinese voice, but only noise.

Start by reproducing the issue with the linked miven model and the reported configuration: no Chinese voice, which produces noise. Inspect the checkpoint and current model around encoder.embedding.weight, compare their [70, 512] and [75, 512] shapes, and confirm that successful Chinese voice generation works without the exception.

Written by the indexing model from the issue text.