Android Speech Recognition is not stable

[juan-serrano-soria]

The continue recording functionality does not work 100% of the times. Sometimes the speech is not correctly added to state when continuing listening.

[yong-asial]

i made some changes to project

• able to run with Monaca (I think you didn't use Monaca to build right?)

• handling the state on Android. I think one of the problem, is there are many timeout to stop the recording every 2-5 seconds - that could be problems. So instead, i implement to stop recording if "listening state" changed or "there is an error in recording".

  SpeechRecognition.addListener('partialResults', (data) => {
      if (continueRecording) {
        setTranscript([...transcript, data.matches[0]]);
      } else {
        setTranscript([data.matches[0]]);
      }
  
      if (Capacitor.platform === 'android') {
          if (!data || !data.matches[0]) {
              stopRecording();
          }
          SpeechRecognition.addListener('listeningState', (state) => {
              if (state && state.status && state.status === 'stopped') {
                  stopRecording();
              }
          });
      }
  
    });
  }

You can import the project from here - https://monaca.mobi/en/directimport?pid=666a6abde7888535448d0191

Anyway, i think the app can be improved like

• request permission should be done only once when the page is finally loaded
• we should keep adding listener and removing listener every times. It is better to just call start and stop.