Han Yang

Well, the pydecoder module is under athena-team project. That is our latest project. Please follow that project https://github.com/athena-team Thanks

Firstly, you should have an acoustic model. Actually, the Athena-Team project have already provide completed examples to build an ASR system: from building acoustic model to decode using or not...

The structure "inner_packed_states_array" is an interface for future expand. Now, for Transformer model, this structure could be ignored because every token records full history labels. So, you could get correct...

It is default setting in Athena-team that \ and \ are the same and are always the last label in table. So the scores of the last dim (3651) is...

> Hey gusy, I finally got some spare time to look into this now. Thanks a lot for putting this together! > > I'm looking at the symbol tables fro...

where do you get the acoustic scores ? We have provide completed examples in https://github.com/athena-team/athena#5-decoding-with-wfst