0

In Chrome I'm recording audio while running speech recognition.
The recording is using AudioRecorder which internally calls getUserMedia.
The speech recognition is using webkitSpeechRecognition.

On the desktop (Win7 + latest Chrome) it works perfectly.
On Android (also latest Chrome) I get an "either or" behavior. When the recognition is working, the recording output is just a quiet file.
When the recording is working (seems random so far) the speech recognition doesn't output any results.
I get no visible error either way

You can see this other question dealing with this combination, although it was probably tested on a desktop and didn't run into the same problem.

Community
  • 1
  • 1
Ron Harlev
  • 16,227
  • 24
  • 89
  • 132

1 Answers1

2

A little more than a year ago doing WebKit speech recognition on android it was possible to do speech recognition and save the audio file from which the recognition was done. We used this to good advantage by sending the audio to other speech recognizers and combining results. I believe it was version 4.0 of android that took that capability away. I think that removal was intentional (for reasons I won't go into here). We still have not found a solution on Android.

FYI: The landscape of Google speech recognition is changing. An important recent change is that repeated speech recognition on a page requires a security certificate. Without it you have to "button click" on the screen to permit each recognition.

industrialpoet
  • 126
  • 7
  • Regardless of the Speech strategy, someone suggested making it expose a connection to getUserMedia so it can share the same audio source http://bit.ly/XQBbtl – Ron Harlev Sep 19 '14 at 17:32