Aime-ASR

aimeasr is a simple JavaScript package for streaming audio to a WebSocket for automatic speech recognition (ASR).

Installation

npm install aimeasr

Use ReactJS friendly

import React, { useEffect, useState } from 'react';
import SpeechRecognition from 'aimeasr';

function App() {
  const [message, setMessage] = useState('');
  const [recognition, setRecognition] = useState(null);

  useEffect(() => {
    async function init() {
      // Custom config
      const speechConfig = SpeechRecognition.defaultConfig;
      speechConfig.sampleRate = 16000;
      speechConfig.speechRecognitionLanguage = "ja-JP";
      speechConfig.authorizationToken = "USER-TOKEN";

      const stream = await navigator.mediaDevices.getUserMedia({
        audio: { echoCancellation: false },
      });
      const recognitionInstance = new SpeechRecognition(stream, speechConfig);
      recognitionInstance.recognizing = (data) => {
        console.log("data:", data);
        if (data.type === 'recognized') {
          setMessage(data.text);
        }
      };
      recognitionInstance.sessionStopped = (s, _e) => {
        console.log("Session stopped event.", s);
      }
      setRecognition(recognitionInstance);
    }
    init();
  }, []);

  return (
    <div>
      <h1>Speech Recognition Demo</h1>
      <button onClick={() => recognition?.start()}>Start</button>
      <button onClick={() => recognition?.stop()}>Stop</button>
      <p>Message: {message}</p>
    </div>
  );
}

export default App;