Unity Voice Recognition Issues

Unity Engine
1

Hello, I’ve been toying with unity’s native window’s voice recognition. I’ve been using the KeywordRecognizer to try to pick up words spoken by the user. The problem is it’s fine if I say the specific keyword alone by itself, but said in a sentence it doesn’t pick it up. For example here’s my script, based on the API:

using System;
using UnityEngine;
using UnityEngine.Windows.Speech;

public class VoiceRecognition : MonoBehaviour
{
    public KeywordTemplate kTemplate;
    private KeywordRecognizer m_Recognizer;
 
 
    void Start()
    {
        m_Recognizer = new KeywordRecognizer(kTemplate.m_Keywords,ConfidenceLevel.Low);
        m_Recognizer.OnPhraseRecognized += OnPhraseRecognized;
        m_Recognizer.Start();
    }

    private void OnPhraseRecognized(PhraseRecognizedEventArgs args)
    {
        StringBuilder builder = new StringBuilder();
        builder.AppendFormat("{0} ({1}){2}", args.text, args.confidence, Environment.NewLine);
        builder.AppendFormat("\tTimestamp: {0}{1}", args.phraseStartTime, Environment.NewLine);
        builder.AppendFormat("\tDuration: {0} seconds{1}", args.phraseDuration.TotalSeconds, Environment.NewLine);
        Debug.Log(builder.ToString());
    }
}

so I have “Hi”, “Hello” and “Ok” as my keywords. Saying them individually is detected fine, but saying into the mic, “Hi There!” or “Hello There!”, “ok i’ll do it” doesn’t detect the word hi, hello or ok respectively. On the other hand, saying “is everything ok?” or “why hello!” works (it detects ok and hello respectively). Any thoughts anyone?

2

I used Google’s speech API after trying out Windows speech recognition and finding it does not work well. Does the Windows built-in translator recognize better than what you are finding from within Unity?

3

According to the Speech-To-Text sample in the asset store, the built in translator is microsoft’s. I decided to try using the dictationrecognizer instead but can’t seem to get it to work properly. It always gives an UnknownError.

4

The DictationCompletionCause.UnknownError received upon request completion seems to be related to your Windows updates. I tried the same piece of code on several devices, with or without updates, and devices which were last updated before October were able to recognize speech.

The problem does not occur when using the 32-bits version of the Unity Editor, or in a 32-bits build; but will always occur on an updated, 64-bits Windows.

“Solution” would be to switch Unity to 32-bits (Editor or only your build configuration); or revert the faulty update (don’t know which one though).

5

Does anyone have an update on this? I’m trying to add a voice-command to a plugin i’m making, and I don’t think I can reasonably ask all of the devs to whom I release it that they should rollback to a 32-bit version of Unity.

6

hwkeyser, this issue seems to be fixed after installing the Windows 10 creators update.

7

hello, im currently running into an issue, im using windows 10 and PhraseRecognitionSystem.isSupported is returning false, does anyone have a clue to why this is happening.