infobot-yandex-stt

infobot-yandex-stt

Node.JS library for Yandex Speech Cloud service. Library can work in stream mode or recognize stored audio file.

To work with this library you need to obtain from Yandex Cloud:

• Private key in PEM format
• Service ID
• Service Key
• Folder ID

Please check this page for information about service accounts.

Audio file recognition example:

var STT = require('infobot-yandex-stt');
var fs = require('fs');

var key = SERVICE_KEY ;
var folder_id = FOLDER_ID;
var service_id = SERVICE_ID;


var stt = new STT(service_id, key, folder_id, fs.readFileSync('./yandex.pem'));
stt.recognizeFile('test.wav').then(res => {
    console.log(JSON.stringify(res));
});

Stream recognition example:

var STT = require('infobot-yandex-stt');
var fs = require('fs');

var key = SERVICE_KEY ;
var folder_id = FOLDER_ID;
var service_id = SERVICE_ID;


var stt = new STT(service_id, key, folder_id, fs.readFileSync('./yandex.pem'));
stt.startRecognitionSession({
        language_code: 'ru-RU', // Possible values: ru-RU, en-US, tr-TR
        sample_rate_hertz: 8000, // Possible values: 8000, 16000, 48000
        audio_encoding: STT.FORMAT_PCM, // Possible values: FORMAT_PCM, FORMAT_OPUS 
        profanity_filter: false, // Make censored output
        partial_results: true // Send partial results
    }).then((recSess) => {
var Writable = require('stream').Writable;
var ws = Writable();
ws._write = function (chunk, enc, next) {
    recSess.writeChunk(chunk);
    next();
};

var readStream = fs.createReadStream(path);
readStream.pipe(ws);

readStream.on("end", function () {
    recSess.finishStream();
});

recSess.on('data', function (data) {
    if (data && data.chunks) {
        console.log(`Transcript: ${data.chunks[0].alternatives[0].text}`);
    }
});

recSess.on('error', function (data) {
    console.error(data);
});
}).catch((err) => {
    console.error(err);
});

Provided by INFOBOT LLC. under Apache 2.0 license.