@vscode/node-speech NPM

node-speech

A node.js binding for a subset of the Azure Speech SDK. Specifically for the Embedded Speech.

The functionality of this module is limited to support the scenarios of the VS Code Speech extension. The goal is not to provide a native binding for the full API of the Azure Speech SDK.

Installation

Install with npm:

npm ci

Usage: Transcription

import * as speech from "@vscode/node-speech";

const modelName = "<name of the speech model>";
const modelPath = "<path to the speech model>";
const modelKey = "<key for the speech model>";

// Live transcription from microphone
let transcriber = speech.createTranscriber(
  { modelName, modelPath, modelKey },
  (err, res) => console.log(err, res)
);
// you can stop/start later
transcriber.stop();
transcriber.start();
// later when done...
transcriber.dispose();

Usage: Synthesizer

import * as speech from "@vscode/node-speech";

const modelName = "<name of the synthesizer model>";
const modelPath = "<path to the synthesizer model>";
const modelKey = "<key for the synthesizer model>";

// Live transcription from microphone
let synthesizer = speech.createSynthesizer(
  { modelName, modelPath, modelKey },
  (err, res) => console.log(err, res)
);
synthesizer.synthesize("Some text to synthesize");
// you can stop later
transcriber.stop();
// later when done...
transcriber.dispose();

Usage: Keyword Recognition

import * as speech from "@vscode/node-speech";

const modelPath = "<path to the keyword model>";

// Keyword recognition from microphone
const result = await speech.recgonize({ modelPath, signal });
console.log(result);