/pocketsphinx-web

Voice recognition in the browser with pocketsphinx

Primary LanguageJavaScriptOtherNOASSERTION

Overview

Pocketsphinx-Web is a an opinionated high-level wrapper for the PocketSphinx.js speech recognizer. It provides a convenient, promise-based API for loading pocketsphinx in a web-worker.

Installing

NPM users can install using:

npm install pocketsphinx-web

Alternatively, you can use a script tag to load pocketsphinx. The api will be available under the objectwindow.Sphinx.

Building

Pocketsphinx-Web is built using webpack. This handles dynamic resolution of worker files and makes including this project in other work much simpler.

yarn install install required packages from the package.json

yarn build-dev - This builds the dev version and stores the bundle under /dev

yarn build - This builds a minified production version of the code with source maps in the /dist folder.

Examples

Web-workers can't be run directly from the filesystem, so to run the examples you will need to create a webserver. You can start one using yarn example, then load the example at localhost:9000. This will also watch for changes to any files and automatically update them for the running example.

Resources

Below are more resources for pocketsphinx and speech recognition in general.