Voice Recognition Stops Listening After a Few Seconds

Web speech API stops listening after some time passes without input

You can listen to the end event and then restart the recognition on the SpeechRecognition object.

You should use a boolean flag for deciding (in the onend event handler), when to restart the recognition (and when not to restart).

You could use the other recognition-related events for this.

E.g. Chrome triggers the following event handlers when recognition is started:

1. onstart
2. onaudiostart

   (only if sound / speech is detected)
3. onsoundstart
4. onspeechstart

If no sound speech is detected, only the first 2 will be triggered, and then, after some timeout, the corresponding end events (in reverse order).

Web speech recognition automatically disconnects

Depending on your browser and if any tabs are trying to access your mic, setting your SpeechRecognition object to continuous should work.

Use the code below for reference. The window.SpeechRecognition is just setting up the speech recognition.

window.SpeechRecognition = window.SpeechRecognition || window.webkitSpeechRecognition;

const recognition = new SpeechRecognition();
recognition.continuous = true;

You can also add an event listener for when your recognition disconnects and start back up the recognition of the web speech api.

recognition.addEventListener('end', () => {
    recognition.start();
  });

This article, from our code world, on the speech api also goes over storing the interim words spoken while it's continuous.

This worked for me on the Chrome browser.

If setting the recognition to continuous does not work, you may have to troubleshoot another service on your computer taking the microphone from your browser tab.

Flutter Speech to text not listening continuously

The Flutter speech to text libraries doesn't allow continuous listening because of privacy issues which is super annoying. The listenFor parameter is an upper bound for when to cut off if you can keep speaking for that long.

As an alternative, I would recommend using Flutter's platform specific implementation. This lets you have a much lower level control. (As a quick tip, even the Android SpeechRecognition library contains a timeout. But the Vosk api is able to do what you're looking for https://github.com/alphacep/vosk-api). If you're really keen, you can look at building your own IME to integrate with voice typing functionality (https://developer.android.com/guide/topics/text/creating-input-method).

---

Related Topics