百度語音測試並導出到Android平臺

參考鏈接:https://www.jianshu.com/p/813b4209458d
新建按鈕控件(改名爲btnRecorder),自動生成canvas和eventsystem,然後canvas下加入text文本控件(不改名)。canvas不用設置,按鈕控件掛載代碼BaiduASR.cs,其他設置如下:
在這裏插入圖片描述
文本控件大小設置如下:
在這裏插入圖片描述
切換到安卓平臺,如下:
在這裏插入圖片描述
player setting如下(包名和.net以及target API版本的設置):
在這裏插入圖片描述
然後導出項目即可。

BaiduASR.cs代碼如下:

using System;
using System.Collections;
using System.Text.RegularExpressions;
using UnityEngine;
using UnityEngine.EventSystems;
using UnityEngine.Networking;
using UnityEngine.UI;

//RequireComponent的這兩個組件主要用於播放自己錄製的聲音,不需要刻意刪除,同時注意刪除使用組件的代碼
[RequireComponent(typeof(AudioListener)), RequireComponent(typeof(AudioSource))]
public class BaiduASR : MonoBehaviour, IPointerDownHandler, IPointerUpHandler
{
//百度語音識別相關key
//string appId = “”;
string apiKey = “bn58jNgdjRED0uKCScW9R2C6”; //填寫自己的apiKey
string secretKey = “LgLYImS2Hcxqno6QQzIMrH7Gf8uRRbZb”; //填寫自己的secretKey

//記錄accesstoken令牌
string accessToken = string.Empty;

//語音識別的結果
string asrResult = string.Empty;

//標記是否有麥克風
private bool isHaveMic = false;

//當前錄音設備名稱
string currentDeviceName = string.Empty;

//錄音頻率,控制錄音質量(8000,16000)
int recordFrequency = 8000;

//上次按下時間戳
double lastPressTimestamp = 0;

//表示錄音的最大時長
int recordMaxLength = 10;

//實際錄音長度(由於unity的錄音需先指定長度,導致識別上傳時候會上傳多餘的無效字節)
//通過該字段,獲取有效錄音長度,上傳時候剪切到無效的字節數據即可
int trueLength = 0;

//存儲錄音的片段
[HideInInspector]
public AudioClip saveAudioClip;

//當前按鈕下的文本
Text textBtn;

//顯示結果的文本
Text textResult;

//音源
AudioSource audioSource;

void Start()
{
	//獲取麥克風設備,判斷是否有麥克風設備
	if (Microphone.devices.Length > 0)
	{
		isHaveMic = true;
		currentDeviceName = Microphone.devices[0];
	}

	//獲取相關組件
	textBtn = this.transform.GetChild(0).GetComponent<Text>();
		audioSource = this.GetComponent<AudioSource>();
			textResult = this.transform.parent.GetChild(1).GetComponent<Text>();
			}

/// <summary>
/// 開始錄音
/// </summary>
/// <param name="isLoop"></param>
/// <param name="lengthSec"></param>
/// <param name="frequency"></param>
/// <returns></returns>
public bool StartRecording(bool isLoop = false) //8000,16000
{
	if (isHaveMic == false || Microphone.IsRecording(currentDeviceName))
	{
		return false;
	}

	//開始錄音
	/*
	 * public static AudioClip Start(string deviceName, bool loop, int lengthSec, int frequency);
	 * deviceName   錄音設備名稱.
	 * loop         如果達到長度,是否繼續記錄
	 * lengthSec    指定錄音的長度.
	 * frequency    音頻採樣率   
	 */

	lastPressTimestamp = GetTimestampOfNowWithMillisecond();

	saveAudioClip = Microphone.Start(currentDeviceName, isLoop, recordMaxLength, recordFrequency);

	return true;
}

/// <summary>
/// 錄音結束,返回實際的錄音時長
/// </summary>
/// <returns></returns>
public int EndRecording()
{
	if (isHaveMic == false || !Microphone.IsRecording(currentDeviceName))
	{
		return 0;
	}

	//結束錄音
	Microphone.End(currentDeviceName);

	//向上取整,避免遺漏錄音末尾
	return Mathf.CeilToInt((float)(GetTimestampOfNowWithMillisecond() - lastPressTimestamp) / 1000f);
}

/// <summary>
/// 獲取毫秒級別的時間戳,用於計算按下錄音時長
/// </summary>
/// <returns></returns>
public double GetTimestampOfNowWithMillisecond()
{
	return (DateTime.Now.ToUniversalTime().Ticks - 621355968000000000) / 10000;
}

/// <summary>
/// 按下錄音按鈕
/// </summary>
/// <param name="eventData"></param>
public void OnPointerDown(PointerEventData eventData)
{
	textBtn.text = "鬆開識別";
	StartRecording();
}

/// <summary>
/// 放開錄音按鈕
/// </summary>
/// <param name="eventData"></param>
public void OnPointerUp(PointerEventData eventData)
{
	textBtn.text = "按住說話";
	trueLength = EndRecording();
	if (trueLength > 1)
	{
		audioSource.PlayOneShot(saveAudioClip);
		StartCoroutine(_StartBaiduYuYin());
	}
	else
	{
		textResult.text = "錄音時長過短";
	}
}

/// <summary>
/// 獲取accessToken請求令牌
/// </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}",
		apiKey, secretKey);
	UnityWebRequest unityWebRequest = UnityWebRequest.Get(uri);
	yield return unityWebRequest.SendWebRequest();

	if (unityWebRequest.isDone)
	{
		//這裏可以考慮用Json,本人比較懶所以用正則匹配出accessToken
		Match match = Regex.Match(unityWebRequest.downloadHandler.text, @"access_token.:.(.*?).,");
		if (match.Success)
		{
			//表示正則匹配到了accessToken
			accessToken = match.Groups[1].ToString();
		}
		else
		{
			textResult.text = "驗證錯誤,獲取AccessToken失敗!!!";
		}
	}
}

/// <summary>
/// 發起語音識別請求
/// </summary>
/// <returns></returns>
IEnumerator _StartBaiduYuYin()
{
	if (string.IsNullOrEmpty(accessToken))
	{
		yield return _GetAccessToken();
	}

	asrResult = string.Empty;

	//處理當前錄音數據爲PCM16
	float[] samples = new float[recordFrequency * trueLength * 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))
	{
		asrResult = unityWebRequest.downloadHandler.text;
		if (Regex.IsMatch(asrResult, @"err_msg.:.success"))
		{
			Match match = Regex.Match(asrResult, "result.:..(.*?)..]");
			if (match.Success)
			{
				asrResult = match.Groups[1].ToString();
			}
		}
		else
		{
			asrResult = "識別結果爲空";
		}
		textResult.text = asrResult;
	}
}

}

發表評論
所有評論
還沒有人評論,想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章