Voice input in Unity

Unity exposes three ways to add Voice input to your Unity application, the first two of which are types of PhraseRecognizer:

• The KeywordRecognizer supplies your app with an array of string commands to listen for
• The GrammarRecognizer gives your app an SRGS file defining a specific grammar to listen for
• The DictationRecognizer lets your app listen for any word and provide the user with a note or other display of their speech

The Microphone capability must be declared for an app to use Voice input.

1. In the Unity Editor, navigate to Edit > Project Settings > Player
2. Select the Windows Store tab
3. In the Publishing Settings > Capabilities section, check the Microphone capability
4. Grant permissions to the app for microphone access on your HoloLens device
   • You'll be asked to do this on device startup, but if you accidentally clicked "no" you can change the permissions in the device settings

To enable your app to listen for specific phrases spoken by the user then take some action, you need to:

1. Specify which phrases to listen for using a KeywordRecognizer or GrammarRecognizer
2. Handle the OnPhraseRecognized event and take action corresponding to the phrase recognized

Namespace: UnityEngine.Windows.Speech
Types: KeywordRecognizer, PhraseRecognizedEventArgs, SpeechError, SpeechSystemStatus

We'll need a few using statements to save some keystrokes:

using UnityEngine.Windows.Speech;
using System.Collections.Generic;
using System.Linq;

Then let's add a few fields to your class to store the recognizer and keyword->action dictionary:

KeywordRecognizer keywordRecognizer;
Dictionary<string, System.Action> keywords = new Dictionary<string, System.Action>();

Now add a keyword to the dictionary, for example in of a Start() method. We're adding the "activate" keyword in this example:

//Create keywords for keyword recognizer
keywords.Add("activate", () =>
{
    // action to be performed when this keyword is spoken
});

Create the keyword recognizer and tell it what we want to recognize:

keywordRecognizer = new KeywordRecognizer(keywords.Keys.ToArray());

Now register for the OnPhraseRecognized event

keywordRecognizer.OnPhraseRecognized += KeywordRecognizer_OnPhraseRecognized;

An example handler is:

private void KeywordRecognizer_OnPhraseRecognized(PhraseRecognizedEventArgs args)
{
    System.Action keywordAction;
    // if the keyword recognized is in our dictionary, call that Action.
    if (keywords.TryGetValue(args.text, out keywordAction))
    {
        keywordAction.Invoke();
    }
}

Finally, start recognizing!

keywordRecognizer.Start();

Namespace: UnityEngine.Windows.Speech
Types: GrammarRecognizer, PhraseRecognizedEventArgs, SpeechError, SpeechSystemStatus

The GrammarRecognizer is used if you're specifying your recognition grammar using SRGS. This can be useful if your app has more than just a few keywords, if you want to recognize more complex phrases, or if you want to easily turn on and off sets of commands. See: Create Grammars Using SRGS XML for file format information.

Once you have your SRGS grammar, and it is in your project in a StreamingAssets folder:

<PROJECT_ROOT>/Assets/StreamingAssets/SRGS/myGrammar.xml

Create a GrammarRecognizer and pass it the path to your SRGS file:

private GrammarRecognizer grammarRecognizer;
grammarRecognizer = new GrammarRecognizer(Application.streamingDataPath + "/SRGS/myGrammar.xml");

Now register for the OnPhraseRecognized event

grammarRecognizer.OnPhraseRecognized += grammarRecognizer_OnPhraseRecognized;

You'll get a callback containing information specified in your SRGS grammar, which you can handle appropriately. Most of the important information will be provided in the semanticMeanings array.

private void Grammar_OnPhraseRecognized(PhraseRecognizedEventArgs args)
{
    SemanticMeaning[] meanings = args.semanticMeanings;
    // do something
}

Finally, start recognizing!

grammarRecognizer.Start();

Namespace: UnityEngine.Windows.Speech
Types: DictationRecognizer, SpeechError, SpeechSystemStatus

Use the DictationRecognizer to convert the user's speech to text. The DictationRecognizer exposes dictation functionality and supports registering and listening for hypothesis and phrase completed events, so you can give feedback to your user both while they speak and afterwards. Start() and Stop() methods respectively enable and disable dictation recognition. Once done with the recognizer, it should be disposed using Dispose() to release the resources it uses. It will release these resources automatically during garbage collection at an extra performance cost if they aren't released before that.

There are only a few steps needed to get started with dictation:

1. Create a new DictationRecognizer
2. Handle Dictation events
3. Start the DictationRecognizer

The Internet Client and Microphone capabilities must be declared for an app to use dictation:

1. In the Unity Editor, go to Edit > Project Settings > Player
2. Select on the Windows Store tab
3. In the Publishing Settings > Capabilities section, check the InternetClient capability
   • Optionally, if you didn't already enable the microphone, check the Microphone capability
4. Grant permissions to the app for microphone access on your HoloLens device if you haven't already
   • You'll be asked to do this on device startup, but if you accidentally clicked "no" you can change the permissions in the device settings

Create a DictationRecognizer like so:

dictationRecognizer = new DictationRecognizer();

There are four dictation events that can be subscribed to and handled to implement dictation behavior.

1. DictationResult
2. DictationComplete
3. DictationHypothesis
4. DictationError

DictationResult

This event is fired after the user pauses, typically at the end of a sentence. The full recognized string is returned here.

First, subscribe to the DictationResult event:

dictationRecognizer.DictationResult += DictationRecognizer_DictationResult;

Then handle the DictationResult callback:

private void DictationRecognizer_DictationResult(string text, ConfidenceLevel confidence)
{
    // do something
}

DictationHypothesis

This event is fired continuously while the user is talking. As the recognizer listens, it provides text of what it's heard so far.

First, subscribe to the DictationHypothesis event:

dictationRecognizer.DictationHypothesis += DictationRecognizer_DictationHypothesis;

Then handle the DictationHypothesis callback:

private void DictationRecognizer_DictationHypothesis(string text)
{
    // do something
}

DictationComplete

This event is fired when the recognizer stops, whether from Stop() being called, a timeout occurring, or some other error.

First, subscribe to the DictationComplete event:

dictationRecognizer.DictationComplete += DictationRecognizer_DictationComplete;

Then handle the DictationComplete callback:

private void DictationRecognizer_DictationComplete(DictationCompletionCause cause)
{
   // do something
}

DictationError

This event is fired when an error occurs.

First, subscribe to the DictationError event:

dictationRecognizer.DictationError += DictationRecognizer_DictationError;

Then handle the DictationError callback:

private void DictationRecognizer_DictationError(string error, int hresult)
{
    // do something
}

Once you've subscribed and handled the dictation events that you care about, start the dictation recognizer to begin receiving events.

dictationRecognizer.Start();

If you no longer want to keep the DictationRecognizer around, you need to unsubscribe from the events and Dispose the DictationRecognizer.

dictationRecognizer.DictationResult -= DictationRecognizer_DictationResult;
dictationRecognizer.DictationComplete -= DictationRecognizer_DictationComplete ;
dictationRecognizer.DictationHypothesis -= DictationRecognizer_DictationHypothesis ;
dictationRecognizer.DictationError -= DictationRecognizer_DictationError ;
dictationRecognizer.Dispose();

Tips

• Start() and Stop() methods respectively enable and disable dictation recognition.
• Once done with the recognizer, it must be disposed using Dispose() to release the resources it uses. It will release these resources automatically during garbage collection at an extra performance cost if they aren't released before that.
• Timeouts occur after a set period of time. You can check for these timeouts in the DictationComplete event. There are two timeouts to be aware of:
  1. If the recognizer starts and doesn't hear any audio for the first five seconds, it will time out.
  2. If the recognizer has given a result, but then hears silence for 20 seconds, it will time out.

If you want to use both phrase recognition and dictation in your app, you'll need to fully shut one down before you can start the other. If you have multiple KeywordRecognizers running, you can shut them all down at once with:

PhraseRecognitionSystem.Shutdown();

You can call Restart() to restore all recognizers to their previous state after the DictationRecognizer has stopped:

PhraseRecognitionSystem.Restart();

You could also just start a KeywordRecognizer, which will restart the PhraseRecognitionSystem as well.

You can find MRTK examples for voice input in the following demo scenes:

• Dictation
• Speech

If you're following the Unity development checkpoint journey we've laid out, you're next task is exploring the Mixed Reality platform capabilities and APIs:

You can always go back to the Unity development checkpoints at any time.