SpeechIOForUnity

-Fri, Aug 28, 2020-

A ready-to-use Unity plugin for Speech Input/Output using native Speech API of both Apple macOS and MS Windows.

github project page is here.

Author: Jotaro Shigeyama and Thijs Roumen

Features

• Custom command speech input
• Support for various speed / language.
• Async/await-based interaction design.

SpeechOut speechOut = new SpeechOut();
SpeechIn  speechIn  = new SpeechIn(OnRecognized);

void Start(){
    Dialog();
}

async void Dialog(){
    await speechOut.Speak("Hello!");
    await speechIn.Listen(new string[] { "Hello", "Hi", "Hey" });
    await speechOut.Speak("How are you doing?");
    await speechIn.Listen(new string[] { "I'm fine", "Nah", "I'm Sick" });
    //...
}

This project repo contains

• Unity project (this repo, tested with v.2019.3.0a8)
• XCode project NSSpeechForUnity (author: Jotaro Shigeyama)
• Visual Studio project WindowsVoiceProject originally from here (https://chadweisshaar.com/blog/2015/07/02/microsoft-speech-for-unity/)
• Unity project contains pre-built .dll and .bundle from above source project.

Installation

This plugin works and tested on macOS Catalina or above, and Windows 10 (Windows 8 is not supported).

OS setup

Right now English / Dutch / German / Japanese are supported. You need to install necessary language module from your OS setting.

Unity

Just simply grab all Scripts, Plugins (and if Scenes if you need) to your own Unity project file.

Potential installation issue for macOS

Some macOS users will experience broken speech input due to missing dictation kits: mostly because of bug on Apple. If you encounter some issue on speech input, please try these.

• make sure your native voice command system works (System Preferences > Accessibility > Voice Command, enabling voice command will invoke macOS system voice command input windows.)
  • Go to “Preferences > Accessibility > Sound Control > Language” and install language pack.
• In case of macOS have some buggy issues: Try switching your OS language to another, then try to install your desired voice command module (macOS will prompt you to download missing dictation model.)
• Make sure your macOS is the latest version
• Try rebooting the system.

dev-Installation

• XCode (mac)
• Visual Studio (win, latest Windows SDK required)

Testing your NSSpeechRecognizer

tested on Apple Swift version 5.1.3

cd NSSpeechTest
swift run

Using in your own Unity package

• Build NSSpeechForUnity in XCode.
• Copy generated .bundle file in Assets/Plugins of your Unity project.

Modifying voice command dictionary

See MyMacSpeechScript.cs.

Documentation

coming soon.

Your feedback is welcome!