文章目录[x]
- 1:简介
- 2:关键字识别
- 3:关键字识别代码
简介
Unity3D为Windows平台提供了语音识别功能，语音识别种类包括：关键字识别、语法识别和听写识别。关键字识别、语法识别可以离线使用，听写识别需要联网。如有离线听写识别的需求可以使用OpenAI开源的whisper项目，在本地部署whisper然后自行添加Web接口。
从Unity 5.4.0开始，Unity引擎添加了windows语音识别API (UnityEngine.Windows.Speech)来识别语音输入。这些api支持所有类型的windows平台(windows editor、windows independent和windows Store)，但只支持windows 10及以上操作系统，支持中文语音识别。
- 操作系统要求： Window10、Windows11 （需要在设置中开启语音识别功能）
- Unity版本:5.4以上
- 引用命名空间： UnityEngine.Windows.Speech
关键字识别
- 1定义要识别的关键字 字符串数组
- 2订阅和处理OnPhraseRecognized事件。当给定短语被识别时，将调用此事件
- 3开始关键字识别
关键字识别代码
为了方便使用，为关键字识别做一个基类WOV(Wake of voice的缩写，我主要用关键字识别后做唤醒操作)
using System;
using System.Collections;
using System.Collections.Generic;
using UnityEngine;
/// <summary>
/// Wake-on-voice 语音唤醒 base类
/// </summary>
public class WOV : MonoBehaviour
{
/// <summary>
/// 关键词回调
/// </summary>
protected Action<string> OnKeywordRecognizer;
/// <summary>
/// 绑定唤醒回调
/// </summary>
/// <param name=""></param>
/// <param name="_callback"></param>
public virtual void OnBindAwakeCallBack(Action<string> _callback)
{
OnKeywordRecognizer += _callback;
}
/// <summary>
/// 开始识别
/// </summary>
public virtual void StartRecognizer()
{
}
/// <summary>
/// 结束识别
/// </summary>
public virtual void StopRecognizer()
{
}
/// <summary>
/// 唤醒词回调
/// </summary>
/// <param name="_msg"></param>
protected virtual void OnAwakeOnVoice(string _msg)
{
if(OnKeywordRecognizer==null)
return;
OnKeywordRecognizer(_msg);
}
}
unity关键字识别类 UnityWakeOnVoice（我主要用关键字识别后做唤醒操作）
using System;
using System.Collections;
using System.Collections.Generic;
using System.Text;
using UnityEngine;
#if UNITY_STANDALONE_WIN
using UnityEngine.Windows.Speech;
using Util;
#endif
/// <summary>
/// unity内置语音唤醒 windows有效
/// </summary>
public class UnityWakeOnVoice : WOV
{
/// <summary>
/// 关键词
/// </summary>
[SerializeField]
private string[] m_Keywords;//要进行关键字识别的数字 如["开灯","关灯","关机"]
/// <summary>
/// 关键字识别器
/// </summary>
#if UNITY_STANDALONE_WIN
private KeywordRecognizer m_Recognizer;
void Start()
{
//创建一个关键字识别器
m_Recognizer = new KeywordRecognizer(m_Keywords,ConfidenceLevel.Medium);
Debug.Log("创建识别器成功");
m_Recognizer.OnPhraseRecognized += OnPhraseRecognized;
StartRecognizer();
}
private void OnDestroy()
{
StopRecognizer();
}
/// <summary>
/// 开始识别
/// </summary>
public override void StartRecognizer()
{
if (m_Recognizer == null)
return;
m_Recognizer.Start();
}
/// <summary>
/// 结束识别
/// </summary>
public override void StopRecognizer()
{
if (m_Recognizer == null)
return;
m_Recognizer.Stop();
}
/// <summary>
/// 识别关键词回调
/// </summary>
/// <param name="args"></param>
private void OnPhraseRecognized(PhraseRecognizedEventArgs args)
{
StringBuilder builder = new StringBuilder();
builder.AppendFormat("{0}", args.text);
string _keyWord = builder.ToString();
Debug.Log("识别器捕捉到关键词："+_keyWord);
OnAwakeOnVoice(_keyWord); //处理回调
KeyWordHandle(_keyWord); //在本类里处理
}
#endif
private void KeyWordHandle(string keyWord)
{
DateTime dateTime = DateTime.Now;
string strNowTime = string.Format("{0:D}-{1:D}-{2:D}-{3:D}-{4:D}-{5:D}-{6:D}", dateTime.Year, dateTime.Month, dateTime.Day, dateTime.Hour, dateTime.Minute, dateTime.Second, dateTime.Millisecond);
Debug.LogWarning($"识别到语音指令：{keyWord}"+" "+strNowTime );
switch (keyWord)
{
//识别不同的关键字后的逻辑
}
}
private void RecordCommandTime(string msg)
{
Debug.Log($"识别到语音指令：{msg}");
}
}