Multi-device Wakeup Arbitration

 

Introduction

Multi-device Wakeup Arbitration (MWA) Engine chooses the most suitable device for a user when multiple devices simultaneously recognize the same trigger word.

It selects the device that can best recognize the user's speech based on the direction, distance, etc. of the user's voice, and responds accordingly.

 

Process of MWA Engine

 

MWA Engine by ThinQ.AI supports the following features:

Feature Description
High accuracy Chooses the most suitable device for the user with exceptional accuracy.
User recognition Recognizes users and provides appropriate feedback.
Server-based operation Requires only a small amount of device resources to use MWA Engine, as user voice commands are processed on the server.
High scalability MWA Engine uses one-channel PCM data so it can be used on most devices.
Efficient network usage Guarantees efficient use of the network by sending only specific utterances to the server.

 

Engine Structure

All features of MWA Engine are run on a server. MWA Engine takes PCM data and JSON data as input and delivers results accordingly.

 

Architecture of MWA Engine

 

Examples of Use

MWA's features can be used in everyday life.

For example, if a user says, "hilg," in a space where home appliances are capable of recognizing the trigger word "hilg," all the devices will recognize the trigger word.

Then, the MWA engine will measure the direction, distance, etc. of the user's voice, select the appliance closest to the user, and provide feedback from the selected appliance.

A picture of appliances with MWA Engines that are close to the user react to the user's keywords