语音数据（Audio Data）

Kinect for windows有四个麦克风，组成一个麦克风阵列，可以采集外部传来的声单，不但可以收集，还可以除噪，识别。Kinect for windows可以识别正前主100度范围的声单。Kinect for windows通过两组数组来定位声单的方向，一个是波束角度（BeamAngle），一个声源角度（SoundSourceAngle）。波束角度是把100度分成10分，分别为50，40，30，20，10，0，-10，-20，-30，-40，-50这11个值，只要在相应的范围内，采样就不会重新进行，比如现在在Kinect for windows的0度方向，大体上在-5或5的范围内，波束角度都不会更新。声源角度就不一样了，只要声音达到一定的分贝，它就采信，同时它还跟随着一个可信度的值，为0~1之间，1是最可信的，就是声音的来源确定性是最可信的。我们可以通过这两个参数结合骨骼定位能就Kinect for windows周围的人进行较为准确的定位。

同时，Kinect对象有一个AudioSource的属性，它有一个Start方法，可以返回录音的数据流。我们对这些数据进行处理就可以。

创建一个WinForm

现在看后台代码：

public partial class Form1 : Form
{
    public Form1()
    {
        InitializeComponent();
        Control.CheckForIllegalCrossThreadCalls = false;
    }

    KinectSensor kinectsensor = null;
    private void Form1_Shown(object sender, EventArgs e)
    {
        //从Kinect集合中找到连接上的Kinect
        foreach (KinectSensor ks in KinectSensor.KinectSensors)
        {
            //找到连接的Kinect
            if (ks.Status == KinectStatus.Connected)
            {
                kinectsensor = ks;
                kinectsensor.Start();//开始工作，即可以采集摄像头和红外摄像头信息
                this.Text = "Kinect开始工作……";

                return;
            }
        }
    }

    private void Form1_FormClosing(object sender, FormClosingEventArgs e)
    {
        if (kinectsensor.Status == KinectStatus.Connected)
        {
            this.kinectsensor.AudioSource.Stop();//结束语音流的采集
            kinectsensor.Stop();//结束Kinect采集工作
            MessageBox.Show("Kinect结束工作！");

        }
    }
    private void AudioReadingThread()
    {
        int second = int.Parse(Second_TB.Text);//Second_TB 中存放着录音的时长，单位为秒
        long len = second * 16000 * 2;//得到一定时间内的文件总长度，采样率是16000，2为采样一次的字节数
        //创建wav文件流，先写文件头，再写声音数据信息
        using (FileStream stream = new FileStream("F:/a/a.wav", FileMode.Create))
        {
            //WAV 文件头
            string wavBegin = "RIFF";
            byte[] tmpbyte = Encoding.Default.GetBytes(wavBegin);
            stream.Write(tmpbyte, 0, tmpbyte.Length);
            // 文件总长度
            long wavLength = len;
            byte[] tmpArr = new byte[4];
            longToIntBinary(wavLength + 16, tmpArr, 0);
            stream.Write(tmpArr, 0, tmpArr.Length);
            // WAV 文件标识
            string wavTag = "WAVEfmt ";
            byte[] tmpbyte1 = Encoding.Default.GetBytes(wavTag);
            stream.Write(tmpbyte1, 0, tmpbyte1.Length);
            // size of .WAV file header
            int headLength = 16;
            tmpArr = new byte[4];
            longToIntBinary(headLength, tmpArr, 0);
            stream.Write(tmpArr, 0, tmpArr.Length);
            // 格式标记 (01 = Windows PCM)
            int wFormatTag = 1;
            tmpArr = new byte[2];
            toShortBinary(wFormatTag, tmpArr, 0);
            stream.Write(tmpArr, 0, tmpArr.Length);
            // 通道数 (1=mono, 2=stereo)
            int nChannels = 1;
            tmpArr = new byte[2];
            toShortBinary(nChannels, tmpArr, 0);
            stream.Write(tmpArr, 0, tmpArr.Length);
            // 采样率
            int nSamplesPerSec = 16000;
            tmpArr = new byte[4];
            longToIntBinary(nSamplesPerSec, tmpArr, 0);
            stream.Write(tmpArr, 0, tmpArr.Length);
            //每秒播放字节数
            int nAvgBytesPerSec = 32000;
            tmpArr = new byte[4];
            longToIntBinary(nAvgBytesPerSec, tmpArr, 0);
            stream.Write(tmpArr, 0, tmpArr.Length);
            // 采样一次占字节数
            int nBlockAlign = 2;
            tmpArr = new byte[2];
            toShortBinary(nBlockAlign, tmpArr, 0);
            stream.Write(tmpArr, 0, tmpArr.Length);// bytes per sample
            // 采样一次占的位数
            int wBitsPerSample = 8 * nBlockAlign;//8乘字节数
            tmpArr = new byte[2];
            toShortBinary(wBitsPerSample, tmpArr, 0);
            stream.Write(tmpArr, 0, tmpArr.Length);// bits per sample

            /** ******以下是数据头********* */
            string dataTag = "data";
            byte[] tmpbyte2 = Encoding.Default.GetBytes(dataTag);
            stream.Write(tmpbyte2, 0, tmpbyte2.Length);// data tag
            //采样数据字节数
            tmpArr = new byte[4];
            longToIntBinary(wavLength, tmpArr, 0);
            stream.Write(tmpArr, 0, tmpArr.Length);// 数据总长度
            //开始采信声音流
            using (Stream audiostream = kinectsensor.AudioSource.Start())
            {
                byte[] buf = new byte[1024];
                int count, totalcount = 0;
                //按1024个字节为一组采集流，并保存，当采信数据没有超过总长度时
                while ((count = audiostream.Read(buf, 0, buf.Length)) > 0 && totalcount < len)
                {
                    stream.Write(buf, 0, count);
                    totalcount += count;
                }
                audiostream.Close();
            }
            stream.Close();
        }
        //让录音按钮生效
        StartRecord_But.Enabled = true;
    }
    //波束角度采集方法
    private void AudioSourceBeamChanged(object sender, BeamAngleChangedEventArgs e)
    {
        double o = e.Angle;
        Beam_Lab.Text = "波束角度：" + o;
    }
    //声源角度和可信度采集方法
    private void AudioSourceSoundSourceAngleChanged(object sender, SoundSourceAngleChangedEventArgs e)
    {
        double o = e.ConfidenceLevel;
        double d = e.Angle;
        Confidence_Lab.Text = "可信度：" + o;
        Sound_Lab.Text = "来源角度：" + d;
    }


    private Thread readingThread;
    /// <summary>
    /// 整型转数组
    /// </summary>
    private static void longToIntBinary(long val, byte[] array, int offset)
    {
        array[offset] = (byte)(val & 0xff);
        array[offset + 1] = (byte)(val >> 8 & 0xff);
        array[offset + 2] = (byte)(val >> 16 & 0xff);
        array[offset + 3] = (byte)(val >> 24 & 0xff);
    }
    /// <summary>
    /// 字节转短整型
    /// </summary>
    private static void byteToShortBinary(byte val, byte[] array, int offset)
    {
        array[offset] = (byte)(val & 0xff);
        array[offset + 1] = 0x0;
    }

    /// <summary>
    /// 将超过0x7FFF的short类型保存为int类型。本方法提供了将有符号short类型转换保存在字节数组中，占据两个字节
    /// </summary>
    private static void toShortBinary(int val, byte[] array, int offset)
    {
        array[offset] = (byte)(val & 0xff);
        array[offset + 1] = (byte)(val >> 8 & 0xff);
    }



    private void StartRecord_But_Click(object sender, EventArgs e)
    {
        //绑定波束角度的委托实例
        kinectsensor.AudioSource.BeamAngleChanged += this.AudioSourceBeamChanged;
        //绑定声音来源角度的委托实例
        kinectsensor.AudioSource.SoundSourceAngleChanged += this.AudioSourceSoundSourceAngleChanged;
        //起一个线程，来完成录音数据的保存
        this.readingThread = new Thread(AudioReadingThread);
        this.readingThread.Start();

        //让录音按钮失效
        StartRecord_But.Enabled = false;
    }

}