search

alphacep / vosk-api

Offline speech recognition API for Android, iOS, Raspberry Pi and servers with Python, Java, C# and Node
Apache License 2.0
7.73k stars 1.08k forks source link

returning two final result #1265

Open marziye-A opened 1 year ago

marziye-A commented 1 year ago

Hi,I encountered a problem while using vosk. why my vosk model returns two final results?one the true transcription and the other is an empty one? any help is really appreciated!

marziye-A commented 1 year ago

does vosk uses silence detection and partitions speech signal accordingly?

nshmyrev commented 1 year ago

does vosk uses silence detection and partitions speech signal accordingly?

Yes

marziye-A commented 1 year ago

can you please elaborate more? how many mili second of non speaking speech is considred as a silence? where can i find the results of different partitions? the final result just returns the result of the last part.

marziye-A commented 1 year ago

does it have anything to do with "n" in "wf.readframes(n)"?

nshmyrev commented 1 year ago

Please check sample code in examples folder