ReferenceError: window is not defined

[sushilbansal]

I tried to use in the following way and got an error:

import React, { useEffect, useState } from 'react';
import SpeechRecognitionPolyfill from 'speech-recognition-aws-polyfill';

const recognition = new SpeechRecognitionPolyfill({
  IdentityPoolId: 'us-west-x:XXX-xxx-xxxxx-xxxxx', // My Identity Pool ID
  region: 'us-west-x', // My AWS region
  lang: 'en-GB'
});

export const AWSTranscribe = () => {
  const [results, setResults] = useState('');
  useEffect(() => {
    recognition.start();
  }, []);
  recognition.onresult = function (event: any) {
    const { transcript } = event.results[0][0];
    setResults(transcript);
    console.log('Heard: ', transcript);
  };

  recognition.onerror = (error: any) => {
    console.log('Somehting went wrong: --- ', error);
  };
  return <div>{results}</div>;
};

[sushilbansal]

@ceuk : probably i am not doing polyfill properly. Can you please suggest?

[sushilbansal]

let me try it out.

[sushilbansal]

yeah the same window issue on latest repo. But i managed to work around that:

const DynamicComponent = dynamic(() => import('./AWSTranscribe'), {
  ssr: false
});

[ceuk]

I'll set up next.js this weekend and recreate/fix the issue

[sushilbansal]

Not sure if it is intended behavior: the audio to text works for 5 seconds and then it does not work. Tested both on create react app and next js in both dev and prod env. With standard SpeechRecognition it start coverting again after a pause. Does not happen in this case.

[ceuk]

It is intentional at the moment yeah until we support continuous