Unity智能语音聊天机器人
在本篇文章中，使用了百度的语音识别、语音合成、智能对话Unit的功能，制作成了一款简易的聊天机器人，在开始做之前呢，需要确定需要实现的核心功能，有以下几点：
（1）实现人机文字聊天
（2）实现人机语音聊天
（3）语音聊天记录播放
（4）文字聊天与语音聊天切换
创建UI界面
核心代码
一、人机文字聊天
chatDialog.onEndEdit.AddListener(delegate
{
if (chatDialog != null)
{
if (chatDialog.text.Equals(""))
{
return;
}
chat.AddChatMessage( ChatUI.enumChatMessageType.MessageRight, chatDialog.text);//自己的聊天UI界面
Unit_NLP(chatDialog.text);
JsonDecode(result);
}
});
/// <summary>
/// NLP的人工智能对话功能
/// </summary>
/// <param name="mysay"></param>
/// <param name="session_id"></param>
/// <param name="action_id"></param>
public void Unit_NLP(string mysay, string session_id = "", string action_id = "")
{
string token = accessToken;
string host = "https://aip.baidubce.com/rpc/2.0/unit/service/chat?access_token=" + token;
HttpWebRequest request = (HttpWebRequest)WebRequest.Create(host);
request.Method = "post";
request.ContentType = "application/json";
request.KeepAlive = true;
JsonData send = new JsonData();
send["version"] = "2.0";
send["service_id"] = "S27034";
send["log_id"] = "home";
send["session_id"] = "home";
send["action_id"] = "1017002";
send["request"] = new JsonData();
send["request"]["user_id"] = "88888";
send["request"]["query"] = mysay;
send["request"]["query_info"] = new JsonData();
send["request"]["query_info"]["type"] = "TEXT";
JsonData bot_session = new JsonData();
bot_session["session_id"] = "";
send["bot_session"] = JsonMapper.ToJson(bot_session);
string sendStr = JsonMapper.ToJson(send);
byte[] buffer = Encoding.UTF8.GetBytes(sendStr);
request.ContentLength = buffer.Length;
request.GetRequestStream().Write(buffer, 0, buffer.Length);
HttpWebResponse response = (HttpWebResponse)request.GetResponse();
StreamReader reader = new StreamReader(response.GetResponseStream(), Encoding.UTF8);
result = reader.ReadToEnd();
}
/// <summary>
/// 对NLP的回复进行Json解析
/// </summary>
/// <param name="js"></param>
/// <returns></returns>
public List<string> JsonDecode(string js)
{
List<string> Says = new List<string>();
var json = SimpleJson.SimpleJson.DeserializeObject<SimpleJson.JsonObject>(js);
if (json.ContainsKey("result"))
{
var result = (SimpleJson.JsonObject)json["result"];
if (result.ContainsKey("response_list"))
{
var resArray = (SimpleJson.JsonArray)result["response_list"];
var res = (SimpleJson.JsonObject)resArray[0];
if (res.ContainsKey("action_list"))
{
var actArray = (SimpleJson.JsonArray)res["action_list"];
var act = (SimpleJson.JsonObject)actArray[0];
if (act.ContainsKey("say"))
{
var say = (string)act["say"];
if (!isChooseSpeech)
{
chat.AddChatMessage(ChatUI.enumChatMessageType.MessageLeft,say);
}
else
{
StartCoroutine(StratTTS(say, s =>
{
robotclipDic.Add(robotindex,s.clip);
robotindex++;
chat.AddChatMessage(ChatUI.enumChatMessageType.MessageLeft);
tts_source.clip = s.clip;
tts_source.Play();
}));
}
Says.Add(say);
}
}
}
}
return Says;
}
二、人机语音聊天
/// <summary>
/// 开始录音
/// </summary>
public void StartRecord()
{
saveAudioClip = Microphone.Start(currentDeviceName, false, recordMaxTime, recordFrequency);
}
/// <summary>
/// 结束录音
/// </summary>
public void EndRecord()
{
Microphone.End(currentDeviceName);
myclipDic.Add(myindex, saveAudioClip);
myindex++;
//source.PlayOneShot(saveAudioClip);
StartCoroutine(RequestASR());//请求语音识别
}
/// <summary>
/// 请求语音识别
/// </summary>
/// <returns></returns>
public IEnumerator RequestASR()
{
if (string.IsNullOrEmpty(accessToken))
{
yield return _GetAccessToken();
}
resulrStr = string.Empty;
//处理当前录音数据为PCM16
float[] samples = new float[recordFrequency * 10 * saveAudioClip.channels];
saveAudioClip.GetData(samples, 0);
var samplesShort = new short[samples.Length];
for (var index = 0; index < samples.Length; index++)
{
samplesShort[index] = (short)(samples[index] * short.MaxValue);
}
byte[] datas = new byte[samplesShort.Length * 2];
Buffer.BlockCopy(samplesShort, 0, datas, 0, datas.Length);
string url = string.Format("{0}?cuid={1}&token={2}", "https://vop.baidu.com/server_api", SystemInfo.deviceUniqueIdentifier, accessToken);
WWWForm wwwForm = new WWWForm();
wwwForm.AddBinaryData("audio", datas);
UnityWebRequest unityWebRequest = UnityWebRequest.Post(url, wwwForm);
unityWebRequest.SetRequestHeader("Content-Type", "audio/pcm;rate=" + recordFrequency);
yield return unityWebRequest.SendWebRequest();
if (string.IsNullOrEmpty(unityWebRequest.error))
{
resulrStr = unityWebRequest.downloadHandler.text;
if (Regex.IsMatch(resulrStr, @"err_msg.:.success"))
{
Match match = Regex.Match(resulrStr, "result.:..(.*?)..]");
if (match.Success)
{
resulrStr = match.Groups[1].ToString();//语音识别的结果
Unit_NLP(resulrStr);
JsonDecode(result);
}
}
else
{
resulrStr = "识别结果为空";
}
}
}
/// <summary>
/// 返回的语音合成结果
/// </summary>
[System.Serializable]
public class TtsResponse
{
public int error_index;
public string error_msg;
public string sn;
public int idx;
public bool Success
{
get {
return error_index == 0; }
}
public AudioClip clip;
}
/// <summary>
/// 请求语音合成
/// </summary>
/// <param name="text"></param>
/// <param name="callback"></param>
/// <returns></returns>
public IEnumerator StratTTS(string text, Action<TtsResponse> callback)
{
var url = "http://tsn.baidu.com/text2audio";
var param = new Dictionary<string, string>();
param.Add("tex", text);
param.Add("tok", accessToken);
param.Add("cuid", SystemInfo.deviceUniqueIdentifier);
param.Add("ctp", "1");
param.Add("lan", "zh");
param.Add("spd", "5");
param.Add("pit", "5");
param.Add("vol", "10");
param.Add("per", "1");
#if UNITY_STANDALONE || UNITY_EDITOR || UNITY_UWP
param.Add("aue", "6"); //Windows设置为wav格式，移动端需要mp3格式
#endif
int i = 0;
foreach (var p in param)
{
url += i != 0 ? "&" : "?";
url += p.Key + "=" + p.Value;
i++;
}
#if UNITY_STANDALONE || UNITY_EDITOR || UNITY_UWP //根据不同平台，获取不同类型的音频格式
var www = UnityWebRequestMultimedia.GetAudioClip(url, AudioType.WAV);
#else
var www = UnityWebRequestMultimedia.GetAudioClip(url, AudioType.MPEG);
#endif
Debug.Log("[WitBaiduAip]" + www.url);
yield return www.SendWebRequest();
if (www.isHttpError || www.isNetworkError)
Debug.LogError(www.error);
else
{
var type = www.GetResponseHeader("Content-Type");
Debug.Log("[WitBaiduAip]response type: " + type);
if (type.Contains("audio"))
{
var response = new TtsResponse {
clip = DownloadHandlerAudioClip.GetContent(www) };
callback(response);
}
else
{
var textBytes = www.downloadHandler.data;
var errorText = Encoding.UTF8.GetString(textBytes);
}
}
}
三、语音聊天记录播放
//发送录音保存字典
public int myindex=0;
public Dictionary<int, AudioClip> myclipDic = new Dictionary<int, AudioClip>();
//收到录音保存字典
public int robotindex = 0;
public Dictionary<int, AudioClip> robotclipDic = new Dictionary<int, AudioClip>();
oid Start()
{
index = UnitManager.instance.myindex;
source = GameObject.Find("Canvas/robot").GetComponent<AudioSource>();
Litsen_btn = transform.GetComponent<Button>();
Litsen_btn.onClick.AddListener(() => {
if (MessageType.Equals("myself"))
{
GetValueByKey(UnitManager.instance.myclipDic, index - 1);
}
else if (MessageType.Equals("robot"))
{
GetValueByKey(UnitManager.instance.robotclipDic, index - 1);
}
});
}
public void GetValueByKey(Dictionary<int, AudioClip> dic, int key)
{
AudioClip clip = null;
dic.TryGetValue(key, out clip);
source.clip = clip;
source.Play();
}
通篇代码
using LitJson;
using System;
using System.Collections;
using System.Collections.Generic;
using System.IO;
using System.Net;
using System.Text;
using System.Text.RegularExpressions;
using UnityEngine;
using UnityEngine.Networking;
using UnityEngine.UI;
public class UnitManager : MonoBehaviour
{
public static UnitManager instance;
private void Awake()
{
instance = this;
}
public string api_key;
public string secret_key;
string accessToken = string.Empty;
InputField chatDialog;
Button speechToggle; //选择当前发送信息的类型（语音或者文字）
public ChatUI chat; //聊天界面
public bool isChooseSpeech = false; //判断当前的发送信息类型状态
//语音识别模块
string resulrStr = string.Empty;
int recordFrequency = 8000;
AudioClip saveAudioClip; //存储语音
string currentDeviceName = string.Empty; //当前录音设备名称
AudioSource source;
int recordMaxTime=20;
public Sprite[] _sp;
public GameObject speechButton;
//NLP
AudioSource tts_source;
string result=string.Empty;
//发送录音保存字典
public int myindex=0;
public Dictionary<int, AudioClip> myclipDic = new Dictionary<int, AudioClip>();
//收到录音保存字典
public int robotindex = 0;
public Dictionary<int, AudioClip> robotclipDic = new Dictionary<int, AudioClip>();
void Start()
{
StartCoroutine(_GetAccessToken());
chat = GameObject.Find("Canvas/ChatUI").GetComponent<ChatUI>();
chatDialog = GameObject.Find("Canvas/ChatUI/InputArea/InputField").GetComponent<InputField>();
speechToggle= GameObject.Find("Canvas/ChatUI/speechToggle").GetComponent<Button>();
tts_source = GameObject.Find("Canvas/ChatUI/speechToggle").GetComponent<AudioSource>();
source = transform.GetComponent<AudioSource>();
chatDialog.onEndEdit.AddListener(delegate
{
if (chatDialog != null)
{
if (chatDialog.text.Equals(""))
{
return;
}
chat.AddChatMessage( ChatUI.enumChatMessageType.MessageRight, chatDialog.text);//自己的聊天UI界面
Unit_NLP(chatDialog.text);
JsonDecode(result);
}
});
speechToggle.onClick.AddListener(ToChangeSpeechToggle);
}
/// <summary>
/// 当前处于发语音还是文字，如果是语音则AI回复语音，如果是文字则AI回复文字
/// </summary>
void ToChangeSpeechToggle()
{
if (!this.isChooseSpeech)
{
this.isChooseSpeech = true;
speechToggle.GetComponent<Image>().sprite = _sp[0];
speechButton.SetActive(true);
}
else
{
this.isChooseSpeech = false;
speechToggle.GetComponent<Image>().sprite = _sp[1];
speechButton.SetActive(false);
}
}
/// <summary>
/// 开始录音
/// </summary>
public void StartRecord()
{
saveAudioClip = Microphone.Start(currentDeviceName, false, recordMaxTime, recordFrequency);
}
/// <summary>
/// 结束录音
/// </summary>
public void EndRecord()
{
Microphone.End(currentDeviceName);
myclipDic.Add(myindex, saveAudioClip);
myindex++;
//source.PlayOneShot(saveAudioClip);
StartCoroutine(RequestASR());//请求语音识别
}
/// <summary>
/// 请求语音识别
/// </summary>
/// <returns></returns>
public IEnumerator RequestASR()
{
if (string.IsNullOrEmpty(accessToken))
{
yield return _GetAccessToken();
}
resulrStr = string.Empty;
//处理当前录音数据为PCM16
float[] samples = new float[recordFrequency * 10 * saveAudioClip.channels];
saveAudioClip.GetData(samples, 0);
var samplesShort = new short[samples.Length];
for (var index = 0; index < samples.Length; index++)
{
samplesShort[index] = (short)(samples[index] * short.MaxValue);
}
byte[] datas = new byte[samplesShort.Length * 2];
Buffer.BlockCopy(samplesShort, 0, datas, 0, datas.Length);
string url = string.Format("{0}?cuid={1}&token={2}", "https://vop.baidu.com/server_api", SystemInfo.deviceUniqueIdentifier, accessToken);
WWWForm wwwForm = new WWWForm();
wwwForm.AddBinaryData("audio", datas);
UnityWebRequest unityWebRequest = UnityWebRequest.Post(url, wwwForm);
unityWebRequest.SetRequestHeader("Content-Type", "audio/pcm;rate=" + recordFrequency);
yield return unityWebRequest.SendWebRequest();
if (string.IsNullOrEmpty(unityWebRequest.error))
{
resulrStr = unityWebRequest.downloadHandler.text;
if (Regex.IsMatch(resulrStr, @"err_msg.:.success"))
{
Match match = Regex.Match(resulrStr, "result.:..(.*?)..]");
if (match.Success)
{
resulrStr = match.Groups[1].ToString();//语音识别的结果
Unit_NLP(resulrStr);
JsonDecode(result);
}
}
else
{
resulrStr = "识别结果为空";
}
}
}
/// <summary>
/// 返回的语音合成结果
/// </summary>
[System.Serializable]
public class TtsResponse
{
public int error_index;
public string error_msg;
public string sn;
public int idx;
public bool Success
{
get {
return error_index == 0; }
}
public AudioClip clip;
}
/// <summary>
/// 请求语音合成
/// </summary>
/// <param name="text"></param>
/// <param name="callback"></param>
/// <returns></returns>
public IEnumerator StratTTS(string text, Action<TtsResponse> callback)
{
var url = "http://tsn.baidu.com/text2audio";
var param = new Dictionary<string, string>();
param.Add("tex", text);
param.Add("tok", accessToken);
param.Add("cuid", SystemInfo.deviceUniqueIdentifier);
param.Add("ctp", "1");
param.Add("lan", "zh");
param.Add("spd", "5");
param.Add("pit", "5");
param.Add("vol", "10");
param.Add("per", "1");
#if UNITY_STANDALONE || UNITY_EDITOR || UNITY_UWP
param.Add("aue", "6"); //Windows设置为wav格式，移动端需要mp3格式
#endif
int i = 0;
foreach (var p in param)
{
url += i != 0 ? "&" : "?";
url += p.Key + "=" + p.Value;
i++;
}
#if UNITY_STANDALONE || UNITY_EDITOR || UNITY_UWP //根据不同平台，获取不同类型的音频格式
var www = UnityWebRequestMultimedia.GetAudioClip(url, AudioType.WAV);
#else
var www = UnityWebRequestMultimedia.GetAudioClip(url, AudioType.MPEG);
#endif
Debug.Log("[WitBaiduAip]" + www.url);
yield return www.SendWebRequest();
if (www.isHttpError || www.isNetworkError)
Debug.LogError(www.error);
else
{
var type = www.GetResponseHeader("Content-Type");
Debug.Log("[WitBaiduAip]response type: " + type);
if (type.Contains("audio"))
{
var response = new TtsResponse {
clip = DownloadHandlerAudioClip.GetContent(www) };
callback(response);
}
else
{
var textBytes = www.downloadHandler.data;
var errorText = Encoding.UTF8.GetString(textBytes);
}
}
}
/// <summary>
/// NLP的人工智能对话功能
/// </summary>
/// <param name="mysay"></param>
/// <param name="session_id"></param>
/// <param name="action_id"></param>
public void Unit_NLP(string mysay, string session_id = "", string action_id = "")
{
string token = accessToken;
string host = "https://aip.baidubce.com/rpc/2.0/unit/service/chat?access_token=" + token;
HttpWebRequest request = (HttpWebRequest)WebRequest.Create(host);
request.Method = "post";
request.ContentType = "application/json";
request.KeepAlive = true;
JsonData send = new JsonData();
send["version"] = "2.0";
send["service_id"] = "S27034";
send["log_id"] = "home";
send["session_id"] = "home";
send["action_id"] = "1017002";
send["request"] = new JsonData();
send["request"]["user_id"] = "88888";
send["request"]["query"] = mysay;
send["request"]["query_info"] = new JsonData();
send["request"]["query_info"]["type"] = "TEXT";
JsonData bot_session = new JsonData();
bot_session["session_id"] = "";
send["bot_session"] = JsonMapper.ToJson(bot_session);
string sendStr = JsonMapper.ToJson(send);
byte[] buffer = Encoding.UTF8.GetBytes(sendStr);
request.ContentLength = buffer.Length;
request.GetRequestStream().Write(buffer, 0, buffer.Length);
HttpWebResponse response = (HttpWebResponse)request.GetResponse();
StreamReader reader = new StreamReader(response.GetResponseStream(), Encoding.UTF8);
result = reader.ReadToEnd();
}
/// <summary>
/// 对NLP的回复进行Json解析
/// </summary>
/// <param name="js"></param>
/// <returns></returns>
public List<string> JsonDecode(string js)
{
List<string> Says = new List<string>();
var json = SimpleJson.SimpleJson.DeserializeObject<SimpleJson.JsonObject>(js);
if (json.ContainsKey("result"))
{
var result = (SimpleJson.JsonObject)json["result"];
if (result.ContainsKey("response_list"))
{
var resArray = (SimpleJson.JsonArray)result["response_list"];
var res = (SimpleJson.JsonObject)resArray[0];
if (res.ContainsKey("action_list"))
{
var actArray = (SimpleJson.JsonArray)res["action_list"];
var act = (SimpleJson.JsonObject)actArray[0];
if (act.ContainsKey("say"))
{
var say = (string)act["say"];
if (!isChooseSpeech)
{
chat.AddChatMessage(ChatUI.enumChatMessageType.MessageLeft,say);
}
else
{
StartCoroutine(StratTTS(say, s =>
{
robotclipDic.Add(robotindex,s.clip);
robotindex++;
chat.AddChatMessage(ChatUI.enumChatMessageType.MessageLeft);
tts_source.clip = s.clip;
tts_source.Play();
}));
}
Says.Add(say);
}
}
}
}
return Says;
}
/// <summary>
/// 获取accessToken请求令牌
/// </summary>
/// <returns></returns>
IEnumerator _GetAccessToken()
{
var uri =
string.Format(
"https://openapi.baidu.com/oauth/2.0/token?grant_type=client_credentials&client_id={0}&client_secret={1}",
api_key, secret_key);
UnityWebRequest unityWebRequest = UnityWebRequest.Get(uri);
yield return unityWebRequest.SendWebRequest();
if (unityWebRequest.isDone)
{
Match match = Regex.Match(unityWebRequest.downloadHandler.text, @"access_token.:.(.*?).,");
if (match.Success)
{
Debug.Log("Token已经匹配");
accessToken = match.Groups[1].ToString();
}
else
{
Debug.Log("验证错误,获取AccessToken失败!!!");
}
}
}
}
演示效果
发送文字：
发送语音：
（因为语音播放演示不了，所以就打个样吧）
结尾，希望能与对这方面感兴趣的兄弟姐妹一起探讨，有需要源码做项目的也可以私聊。技术有限，现在是在Windows平台上实现了简易智能语音聊天机器人的基本功能，各方面的细节还有待优化，本篇结束啦，希望明天会更好。