Saturday, 3 June 2017

Why is the API response time for Audio Stream Recognition so slow?

I am using very similar code to Google's example for performing speech recognition on an Audio stream using the Node.js client library.

The API is parsing my audio correctly, but I find myself waiting 30-45 seconds before I get a response. Considering how snappy the demo is, this doesn't seem right. Is there something I am configuring incorrectly on my end?

I've tried writing to a local file instead just to make sure the audio is coming through clearly, and the recording seems fine.

Thanks for any help you can give!

import record from 'node-record-lpcm16';
import Speech from '@google-cloud/speech';


function streamToParser(){
  const speech = Speech();
  const request = {
    config: {
      encoding: 'LINEAR16',
      sampleRateHertz: 16000,
      languageCode: 'en-US',
    },
    interimResults: true,
  };

  const recognizeStream = speech.createRecognizeStream(request)
  .on('error', console.error)
  .on('data', (data) => {
    console.log(data.results)
  });

  record
  .start({
    sampleRate: 16000,
    threshold: .6,
    verbose: true,
    silence: '5.0'
  })
  .on('error', console.error)
  .pipe(recognizeStream)

  console.log('Google is listening...')
};



streamToParser();



via JWayne

No comments:

Post a Comment