How to replicate result for AISHELL3 demo result

[zhaoyukoon]

Hi~ I tried to download pretrained AISHELL3 model and run synthesize with different speaker id. However, the generated wav file sounds different.

1. What is the relation between speaker_id and id in AISHELL3 dataset like SSB1699?
2. What is the param for replicating same wav result as in demo/AISHELL3?

Thanks a lot!