quill-stt

Quill Speech-to-Text Module

A modern speech recognition module for the Quill rich text editor that adds speech input capabilities with a moderately beautiful UI.

Features

• 🎤 Native browser Speech Recognition API support
• 📝 Insert speech as text at cursor position
• 📊 Moderately beautiful audio visualization during recording with customizable colors
• 🔌 Extensible with custom speech recognition processors
• 🌐 Support for multiple languages
• 🔄 Continuous recording mode option
• 🎨 Modern and responsive UI

Installation

npm install quill-stt --save
# or
yarn add quill-stt

Usage

// Import Quill and the module
import Quill from 'quill';
import 'quill-stt';

// Initialize Quill with the module
const quill = new Quill('#editor', {
  theme: 'snow',
  modules: {
    toolbar: [
      ['bold', 'italic', 'underline'],
      ['image', 'code-block']
    ],
    speechToText: {
      language: 'en-US',
      continuous: false,
      visualizer: true,
      waveformColor: '#4285f4', 
      histogramColor: '#25D366' 
    }
  }
});

Configuration Options

The speech-to-text module accepts the following options:

Option	Type	Default	Description
position	String	'bottom'	Position of the speech input toolbar. Can be 'bottom' or 'top'.
language	String	'en-US'	Language for speech recognition. Use BCP 47 language tags like 'en-US', 'fr-FR', 'es-ES'.
continuous	Boolean	false	Whether to keep recording after results are finalized.
visualizer	Boolean	true	Show audio visualizer during recording.
waveformColor	String	'#4285f4'	Color of the waveform line in the audio visualizer. Can be any valid CSS color.
histogramColor	String	'#25D366'	Color of the histogram bars in the audio visualizer. Can be any valid CSS color.
interimResults	Boolean	true	Show interim results during speech recognition.
customProcessor	Object	null	Custom processor for speech recognition (see below).

Events

You can subscribe to events from the speech-to-text module:

const speechToText = quill.getModule('speechToText');

speechToText.on('onStart', () => {
  console.log('Recording started');
});

speechToText.on('onResult', (data) => {
  console.log(`Received ${data.isFinal ? 'final' : 'interim'} result: "${data.transcript}"`);
});

speechToText.on('onEnd', () => {
  console.log('Recording stopped');
});

speechToText.on('onError', (error) => {
  console.error('Speech recognition error:', error);
});

Using a Custom Speech Processor

If you want to use a custom speech recognition service (like OpenAI's Whisper API), you can provide a custom processor:

const quill = new Quill('#editor', {
  // ...other options
  modules: {
    // ...other modules
    speechToText: {
      customProcessor: {
        // Will be called when recording starts
        start() {
          // Your code to start audio recording
          // For example, connect to WebSocket for streaming to a Whisper API endpoint
        },
        
        // Will be called when recording stops
        stop() {
          // Your code to stop audio recording
        },
        
        // These will be set by the module
        onResult: null, // Call this with (transcript, isFinal) when results are available
        onEnd: null,    // Call this when processing ends
        onError: null   // Call this with error object when an error occurs
      }
    }
  }
});

Browser Compatibility

This module uses the Web Speech API, which is supported in most modern browsers like Chrome, Edge, Safari, and Firefox. See browser compatibility for more details.

If the Speech Recognition API is not available in the browser, you can provide a custom speech processor to use a different service.

Development

To run the development server:

npm start

To build for production:

npm run build

License

MIT