trying to do speech recognition for a very simple language

I'm trying to do speech recognition for a very simple language.
I am trying to build it with my own data based on the vosk-api/training script.

Building the model seems to be going well, but it doesn't work well in Vosk.
How should I structure the files?

Attach the modified run.sh and the build log.
Thank you.

[run.sh.txt](https://github.com/alphacep/vosk-api/files/14367516/run.sh.txt)
[run.log](https://github.com/alphacep/vosk-api/files/14367517/run.log)