Wrong syllable analysis for "silence" and "silent"

When using the compromise-syllables plugin, it counts the amount of syllables in the word “silent” as 1, which is incorrect:

import compromise from 'compromise';
import syllables from 'compromise-syllables';

const nlp = compromise.extend(syllables);

const word = 'silent';

console.log(nlp(word).terms().syllables()[0].syllables);
// ['silent']

// should be: ['si', 'lent']

The same is true for silence (['silen', 'ce']), silenced ([ 'silen', 'ced' ]) and other similar words.

I suspect there are other words affected by this.

[spencermountain]

i mean, the best would be if it could be configured somehow, for different dialects. Having a configurable way to specify syllable pronunciation - that would be so fun to play with. the compromise tokenizer lends itself really well to this - and I think a low-data solution for the browser could become really popular, for a lot of uses.

[spencermountain]

thank you @axaysushir - go crazy! you’ll find the very-weak syllable function here - it would be really fun to make this good.

so my understanding is that a proper syllable function will first do pronounciation step - which you can see a early implementation of here - combining these somehow would be really fun, and have a lot of applications.

happy to help-out, in any way, cheers