simple-speech-recognition
v1.0.3
Published
A simple wrapper for Speech Recognition APIs in the browser
Downloads
6
Maintainers
Readme
simple-speech-recognition
Why make it complicated if it can be simple?
This small library abstracts away all those nitty-gritty details you have to pay attention to when using the
browser-integrated SpeechRecognition
, and lets you focus on what's important: your own code.
Speech Recognition
Speech recognition is an experimental technology, which is currently being implemented in Browsers (at the time of writing, Chrome is the only one to fully support this).
With this technology you can put in texts into your computer using voice: https://developer.mozilla.org/en-US/docs/Web/API/SpeechRecognition
This library is a wrapper around the in-browser API. It "fixes" some of the quirky behaviour I've encountered and makes it more simple to be used.
Installation
Install this package using npm
or yarn
:
npm install simple-speech-recognition
yarn add simple-speech-recognition
Usage
Import the SpeechRecognizer
class:
import SpeechRecognizer from 'simple-speech-recognition'
Create one (or more!) instance from the imported class:
const speechRecognizer = new SpeechRecognizer(options)
options
can consist of the following fields:
const options = {
SpeechRecognition: window.SpeechRecognition || window.webkitSpeechRecognition || DummySpeechRecognition,
timeout: 1000, // The timeout until a speech recognition is completed (after the user has spoken their last word)
resetCallback, // Callback used whenever an error happens
resultCallback, // Callback used for results
lang: 'en-US', // Language set on the Speech Recognition object
interimResults: true // Whether you want to receive interim results or not
}
Simply pass an object with any fields you want to override.
The resultCallback
receives the following message:
{
transcript: 'Lorem ipsum dolor sit amet', // The recognized words
finished: false // Whether the received message is the last one in the current recognition cycle
}
You won't receive any messages with finished: false
if you passed interimResults: false
to the constructor.
Example
Here is a simple example, which will log speech recognition results to the console:
import SpeechRecognizer from 'simple-speech-recognition'
const speechRecognizer = new SpeechRecognizer({
resultCallback: ({ transcript, finished }) => console.log(transcript)
})
speechRecognizer.start()