1
votes

I am working on google cloud speech api i tried their api with my flac file but they are just transcripting a some part of my flac format file.Is there anything i can do with my file to make it work perfectly.I tried testing their api on their link.I Pass this in the request body

{ "audio": { "uri": "gs://kami1994/file-1506682082651.flac" }, "config": { "encoding": "FLAC", "languageCode": "en-US", "sampleRateHertz": 44100, "maxAlternatives": 1 } }

But it is not returning text exactly like i spoke.

1

1 Answers

0
votes

Speech recognition engine is expected to be highly accurate but not 100%. To debug your problem further you can enable word time offsets, in recognize config.

"config": { "encoding": "FLAC", "languageCode": "en-US", "sampleRateHertz": 44100, "maxAlternatives": 1, **"enable_word_time_offsets": True** }

This would help you check if the word boundaries are detected.