@readapt/text-engine

Installation

npm

1. Install @readapt/text-engine package:

     npm install @readapt/text-engine

Usage

Directly in your browser add script

<script src="path/to/yourCopyOf/readapt-text-engine.browser.js"></script>

This script add the variable textEngine on the browser global scope.

node (babel, typescript)

Import entire module

import * as TextEngine from '@readapt/text-engine'

or individually

import { analyse, getPhonemes } from '@readapt/text-engine'

node (cjs)

Entire module

const textEngine = require('@readapt/text-engine')

or

const { analyse, getPhonemes } = require('@readapt/text-engine')

Text Analyse

For example analyse the sentence salut les amies:

const textAnalysis = analyse('salut les amies!', 'fr')

Output:
{
  phonemes:      [  31, 1, 19, 33,  0, 0, 19, 8, 0, 0, 1, 20, 15, 0, 0, 0, 0],
  syllables:     [ [ 0, 1 ], [ 2, 4 ], [ 6, 8 ], [ 10, 10 ], [ 11, 14 ] ],
  silentLetters: [ 4, 8, 13, 14 ],
  liaison:       [ [ 8, 10, 'z' ] ] // only when lang is fr
}

Get fr language phonemes list

const frPhonemes = getPhonemes('fr')

// Output
// [ '_',  'a', 'an', 'b',  'c', 'ch', ... , 'w',  'y',   'z',  'iy', 'tch', 'dg']

Then you can map the phonemes of the sentence with the phonemes of the lang

const phonemes = textAnalysis.phonemes.map(index => frPhonemes[index])

// Output
// [ 's', 'a', 'l', 'u', '_', '_', 'l', 'é', '_', '_', 'a', 'm', 'i', '_', '_', '_']

const textPhonemes = phonemes.join(' ')

// Output
// s a l u _ _ l é _ _ a m i _ _ _

To determine full pronunciation you need to add the liaisons if they are present

// salu lé ami
//       z‿
// with liaisons

In English the analysis is very similar but without the field liaisons.

Some words are more than one phoneme in this case the phonemes array may be contain a tuple of phonemes.

For example analyse the sentence John was subtle:

const textAnalysisEn = analyse('John was subtle', 'en')

Output:
{
  phonemes:      [ 19, 1, 0, 23, 0, 36, 1, 38, 0, 29, 3, 3, 31, [3, 21], 0 ],
  syllables:     [ [0, 3], [5, 7], [9, 11], [12, 14] ],
  silentLetters: [2, 11]
}

The sentence above, the grapheme l has two phonemes ('ah', and 'l'). In this case the mapping phoneme index to phoneme is a bit different because you need to call flat() first.

const enPhonemes = getPhonemes('en')

const phonemes = textAnalysisEn.phonemes.flat().map((index => enPhonemes[index]).join(' ')

// Output:
// jh aa _ n _ w aa z _ s ah ah t ah l _'