Welcome to A.V.A.T.A.R

A.V.A.T.A.R is an open source client-server speech recognition assistant dedicated to the design of intelligent homes and the Internet of Things.

A.V.A.T.A.R was developed using Electron framework and Node.js.

Principle of use

The principle is to dictate a rule via a client, which is sent to the server. The server translates it into English and analyzes it using Natural Language Processing (NLP).
If an intention is found, the server triggers the associated script. These scripts (or plugins) developed in Node.js interact with all connected objects that can be controlled (home automation box, TV box, TV, Wi-Fi speaker, Sonos, lamp, gadget...), Open Data (TV programs, cinema schedules, world weather, wikipedia...) or even other voice assistants (like Google assistant)

A.V.A.T.A.R is the “core” of your applications, so all you have to do is develop plugins that interact with what you want to drive, and A.V.A.T.A.R takes care of the rest. To this end, A.V.A.T.A.R offers you a development API and a host of tools for easily creating and managing your plugins, as well as a library of plugins made by the community of contributors.

If you want to dispense with voice recognition, A.V.A.T.A.R also offers a tool for creating and managing button widgets in server and client interfaces. These widgets also interact with home automation boxes.

You can also add your own JS/HTML/CSS windows to the application.

Supported platforms

System	Version	arch	Comment
Windows	>= 10	ia32 (x86) x64 (amd64) arm64	Voices available through system voice packages and Chrome speech synthesis
linux	Debian Fedora Ubuntu	x64 (amd64) arm64	Available voices by `espeak` and `mbrola`
mac0S	>= Sonoma	x64 (amd64) arm64	Voices available through system voice packages and Chrome speech synthesis
Raspberry Pi	Raspberry Pi OS	arm64	Only for A.V.A.T.A.R Server. No voice recognition available, no voice available for clients.

A.V.A.T.A.R was developed on Windows 10-11, macOS Sonoma and linux (Debian 12).