{"status":"ok","data":[{"confidence":0.932974,"text":""}]}

[zhaoyi2]

Hi,it still don't work well with the latest kaldi,so anyone can help?

[formigone]

@zhaoyi2 I was having the same issue. I was using asr-server from this repo which uses a fork of this repo.

My solution was to use the following to send audio to the server (using the demo model mentioned in the docs):

curl -H "Content-Type: application/octet-stream" --data-binary @/Users/my-user/Downloads/sound-file.wav http://localhost:8080/asr?endofspeech=1

In other words, I played around with the parameters specified in the docs. With the endofspeec=1 param, I now get the entire file transcribed (albeit, the sample model isn't super accurate with the input files I'm testing this with).