Continuous Mode Speech Recognition Timing Out

0 favourites
  • 4 posts
From the Asset Store
Whisper is a speech recognition system for making requests to the OpenAI speech to text API
  • Problem Description

    Hi - I don't know whether this a feature or a bug but Speech Recognition in continuous mode stops working if the player doesn't say anything for a while (after they have previously been talking) despite the fact that the manual states it should continue until the page is closed. When it stops recognising speech in this scenario no error is passed to SpeechError so I'm unsure why it is stopping.

    Attach a Capx

    https://dl.dropboxusercontent.com/u/31546757/speechrecogbugreport.capx

    Description of Capx

    Starts continuous mode speech recognition when the player hits start and reports whether speech is recognised and the content of SpeechError.

    Steps to Reproduce Bug

    • Click on 'Start' to begin speech recognition
    • Say a bunch of stuff into the mic
    • Stay quiet for a while

    Observed Result

    Speech recognition stops after the player hasn't spoken for a while and no error is passed to SpeechError

    Expected Result

    Speech recognition continues until the page is closed

    Affected Browsers

    • Chrome: YES
    • FireFox: NOT APPLICABLE
    • Internet Explorer: NOT APPLICABLE

    Operating System and Service Pack

    Windows 7 Home Premium SP1

    Construct 2 Version ID

    221 (64 bit)

  • I believe this is how the browsers intend for speech recognition to work, rather than it being a C2 bug. If you don't speak for a while the browser stops it with a "no-speech" error. I guess you could just start speech recognition again?

  • Try Construct 3

    Develop games in your browser. Powerful, performant & highly capable.

    Try Now Construct 3 users don't see these ads
  • Thanks for the response Ashley - I thought it might be a browser issue.

    The only thing is that no error is passed to SpeechError in the specific scenario of the player not speaking for while after having spoken previously. If you start speech recognition and say nothing at all 'no-speech' is indeed passed to SpeechError. However, if you start it, say something, then stay quiet, no error is passed even though it stops working. Unsure whether that's the browser not reporting it or C2 not picking it up.

    Yeah, my workaround is to indicate in game that the speech recognition has stopped and prompt the player to reinitiate it. The only vaguely annoying thing about this is it causes Chrome to ask for access to the mic again which is a rather immersion breaking.

  • will it works as a mobile app??

Jump to:
Active Users
There are 1 visitors browsing this topic (0 users and 1 guests)