Overview

The Vuzix Speech Command engine is a fully embedded, fast, phrase-matching recognition system designed to interpret and respond to voice commands. A platform based vocabulary is available to all apps; it is intended to facilitate default navigation and selection without direction from the client app. That is, a client app can benefit from navigation provided by a base vocabulary with no setup or explicit awareness of the speech command engine. This capability is implemented by mapping phrases to Android key events.

For many applications, it is desirable to implement a custom vocabulary which performs application-specific actions when an application-specific phrase is spoken (e.g. to capture a still image when “take a picture” is spoken.) The Vuzix Speech Command engine provides two mechanisms by which this can be achieved: Android key events and Android intents.

Custom Vocabulary Architecture

The Vuzix Speech Command engine is implemented as an Android service that runs locally on the device. No cloud servers are used, and the audio data never leaves the device.

Each Activity can have its own vocabulary. The system will automatically switch to the proper vocabulary as Activities are paused and resumed. If no vocabulary is provided, the system will use the default navigation commands.

A 3rd party Application may create a custom vocabulary for the Vuzix Speech Command engine by utilizing the Vuzix Speech SDK as described throughout this section of the knowledge base.

Low Code Approach

Vuzix Speech commands can be partially leveraged without deploying the SDK into your application by adding XML tags in your application’s views. This would be particularly interesting to applications that are maintained for use on more devices than just Vuzix devices.
Please see the Low Code Approach article for more information.