Closed ghost closed 2 years ago
Hi, u can set language_type=1
to preserve blanks and lowercacse=false
to capitalize english words.
https://github.com/wenet-e2e/wenet/blob/main/runtime/core/decoder/params.h#L63-L68
https://github.com/wenet-e2e/wenet/blob/main/runtime/core/post_processor/post_processor.cc#L26-L56
Bug Detail Hello, I'm an engineer who makes a voice recognition model through your wenet library. For your honor, we successfully made our e2e model with reasonable WER. As a result of using various decoding methods, there was no problem with the decoding functions implemented in python. However, when using WFST decoding method which is compiled in runtime C++ code, it made an issue.
The database we use is a mixture of English, Korean, and numbers, and as a result of WFST decoding, all space disappears in languages other than English. Also, there was a phenomenon that only lowercase letters appeared in English without capitalization and lowercase letters.
Bug Result
Is there any solution to a problem like this?