Implement the Connectionist Temporal Classification(CTC) beam search in JavaScript. The input is log probabilities of an array. The length of the array is the number of CTC slots. Each item in the array contains an array of log probabilities to each characters, including blank character. Usually the blank character is the last one. The implementation doesn't support NGram now. But it's one of the todos.
The following code is used to handle English CTC results:
const { CTCBeamSearch, EN_VOCABULARY } = require('ctc-beam-search');
const bs = new CTCBeamSearch(EN_VOCABULARY);
const data = ....; // log probabilities
const results = bs.search(data, 5); // beam width = 5
// dump the first result to console as a string
console.log(results[0].convertToStr(EN_VOCABULARY));
The EN_VOCABULARY
is like this:
const { Vocabulary } = require('ctc-beam-search');
const engV = new Vocabulary({ ' ': 0,
'a': 1,
'b': 2,
'c': 3,
'd': 4,
'e': 5,
'f': 6,
'g': 7,
'h': 8,
'i': 9,
'j': 10,
'k': 11,
'l': 12,
'm': 13,
'n': 14,
'o': 15,
'p': 16,
'q': 17,
'r': 18,
's': 19,
't': 20,
'u': 21,
'v': 22,
'w': 23,
'x': 24,
'y': 25,
'z': 26,
'\'': 27,
}, 28);
You can create you own Vocabulary.