此系列文章拖了N久，有好多人发邮件来询问我第五次的文章为什么没有写，其实非常抱歉，本人学生一个，暑假一直
去公司实习，最近又忙着各种招聘找工作，没有时间好好写，现在抽空把最后一篇补上，水平有限，如过有不对的，请
各位指正～
前四篇文章分别介绍了 “代码结构”，“程序流程”，以及”发送方的处理”，现在就把接收方的处理流程做个介绍；
如上图所示，接收方的操作有三个类：AudioDecoder(负责解码），AudioPlayer(负责播放解码后的音频），
AudioReceiver(负责从服务器接收音频数据包），这三个类的流程在第三篇中有详细的介绍。
1.AudioReceiver代码：
AudioReceiver使用UDP方式从服务端接收音频数据，其过程比较简单，直接上代码：
package xmu.swordbearer.audio.receiver;import java.io.IOException;import java.net.DatagramPacket;import java.net.DatagramSocket;import java.net.SocketException;import xmu.swordbearer.audio.MyConfig;import android.util.Log;public class AudioReceiver implements Runnable { String LOG = "NET Reciever "; int port = MyConfig.CLIENT_PORT;// 接收的端口 DatagramSocket socket; DatagramPacket packet; boolean isRunning = false; private byte[] packetBuf = new byte[1024]; private int packetSize = 1024; /* * 开始接收数据 */ public void startRecieving() { if (socket == null) { try { socket = new DatagramSocket(port); packet = new DatagramPacket(packetBuf, packetSize); } catch (SocketException e) { } } new Thread(this).start(); } /* * 停止接收数据 */ public void stopRecieving() { isRunning = false; } /* * 释放资源 */ private void release() { if (packet != null) { packet = null; } if (socket != null) { socket.close(); socket = null; } } public void run() { // 在接收前，要先启动解码器 AudioDecoder decoder = AudioDecoder.getInstance(); decoder.startDecoding(); isRunning = true; try { while (isRunning) { socket.receive(packet); // 每接收一个UDP包，就交给解码器，等待解码 decoder.addData(packet.getData(), packet.getLength()); } } catch (IOException e) { Log.e(LOG, LOG + "RECIEVE ERROR!"); } // 接收完成，停止解码器，释放资源 decoder.stopDecoding(); release(); Log.e(LOG, LOG + "stop recieving"); }}2.AudioDecoder代码：
解码的过程也很简单，由于接收端接收到了音频数据，然后就把数据交给解码器，所以解码的主要工作就是把接收端的数
据取出来进行解码，如果解码正确，就将解码后的数据再转交给AudioPlayer去播放，这三个类之间是依次传递的 :
AudioReceiver---->AudioDecoder--->AudioPlayer
下面代码中有个List变量 private List<AudioData> dataList = null;这个就是用来存放数据的，每次解码时，dataList.remove(0)，
从最前端取出数据进行解码：
package xmu.swordbearer.audio.receiver;import java.util.Collections;import java.util.LinkedList;import java.util.List;import xmu.swordbearer.audio.AudioCodec;import xmu.swordbearer.audio.data.AudioData;import android.util.Log;public class AudioDecoder implements Runnable { String LOG = "CODEC Decoder "; private static AudioDecoder decoder; private static final int MAX_BUFFER_SIZE = 2048; private byte[] decodedData = new byte[1024];// data of decoded private boolean isDecoding = false; private List<AudioData> dataList = null; public static AudioDecoder getInstance() { if (decoder == null) { decoder = new AudioDecoder(); } return decoder; } private AudioDecoder() { this.dataList = Collections .synchronizedList(new LinkedList<AudioData>()); } /* * add Data to be decoded * * @ data:the data recieved from server * * @ size:data size */ public void addData(byte[] data, int size) { AudioData adata = new AudioData(); adata.setSize(size); byte[] tempData = new byte[size]; System.arraycopy(data, 0, tempData, 0, size); adata.setRealData(tempData); dataList.add(adata); System.out.println(LOG + "add data once"); } /* * start decode AMR data */ public void startDecoding() { System.out.println(LOG + "start decoder"); if (isDecoding) { return; } new Thread(this).start(); } public void run() { // start player first AudioPlayer player = AudioPlayer.getInstance(); player.startPlaying(); // this.isDecoding = true; // init ILBC parameter:30 ,20, 15 AudioCodec.audio_codec_init(30); Log.d(LOG, LOG + "initialized decoder"); int decodeSize = 0; while (isDecoding) { while (dataList.size() > 0) { AudioData encodedData = dataList.remove(0); decodedData = new byte[MAX_BUFFER_SIZE]; byte[] data = encodedData.getRealData(); // decodeSize = AudioCodec.audio_decode(data, 0, encodedData.getSize(), decodedData, 0); if (decodeSize > 0) { // add decoded audio to player player.addData(decodedData, decodeSize); // clear data decodedData = new byte[decodedData.length]; } } } System.out.println(LOG + "stop decoder"); // stop playback audio player.stopPlaying(); } public void stopDecoding() { this.isDecoding = false; }}3.AudioPlayer代码：
播放器的工作流程其实和解码器一模一样，都是启动一个线程，然后不断从自己的 dataList中提取数据。
不过要注意，播放器的一些参数配置非常的关键；
播放声音时，使用了Android自带的 AudioTrack 这个类，它有这个方法：
public int write(byte[] audioData,int offsetInBytes, int sizeInBytes)可以直接播放；
所有播放器的代码如下：
package xmu.swordbearer.audio.receiver;import java.util.Collections;import java.util.LinkedList;import java.util.List;import xmu.swordbearer.audio.data.AudioData;import android.media.AudioFormat;import android.media.AudioManager;import android.media.AudioRecord;import android.media.AudioTrack;import android.util.Log;public class AudioPlayer implements Runnable { String LOG = "AudioPlayer "; private static AudioPlayer player; private List<AudioData> dataList = null; private AudioData playData; private boolean isPlaying = false; private AudioTrack audioTrack; private static final int sampleRate = 8000; // 注意：参数配置 private static final int channelConfig = AudioFormat.CHANNEL_IN_MONO; private static final int audioFormat = AudioFormat.ENCODING_PCM_16BIT; private AudioPlayer() { dataList = Collections.synchronizedList(new LinkedList<AudioData>()); } public static AudioPlayer getInstance() { if (player == null) { player = new AudioPlayer(); } return player; } public void addData(byte[] rawData, int size) { AudioData decodedData = new AudioData(); decodedData.setSize(size); byte[] tempData = new byte[size]; System.arraycopy(rawData, 0, tempData, 0, size); decodedData.setRealData(tempData); dataList.add(decodedData); } /* * init Player parameters */ private boolean initAudioTrack() { int bufferSize = AudioRecord.getMinBufferSize(sampleRate, channelConfig, audioFormat); if (bufferSize < 0) { Log.e(LOG, LOG + "initialize error!"); return false; } audioTrack = new AudioTrack(AudioManager.STREAM_MUSIC, sampleRate, channelConfig, audioFormat, bufferSize, AudioTrack.MODE_STREAM); // set volume:设置播放音量 audioTrack.setStereoVolume(1.0f, 1.0f); audioTrack.play(); return true; } private void playFromList() { while (dataList.size() > 0 && isPlaying) { playData = dataList.remove(0); audioTrack.write(playData.getRealData(), 0, playData.getSize()); } } public void startPlaying() { if (isPlaying) { return; } new Thread(this).start(); } public void run() { this.isPlaying = true; if (!initAudioTrack()) { Log.e(LOG, LOG + "initialized player error!"); return; } while (isPlaying) { if (dataList.size() > 0) { playFromList(); } else { try { Thread.sleep(20); } catch (InterruptedException e) { } } } if (this.audioTrack != null) { if (this.audioTrack.getPlayState() == AudioTrack.PLAYSTATE_PLAYING) { this.audioTrack.stop(); this.audioTrack.release(); } } Log.d(LOG, LOG + "end playing"); } public void stopPlaying() { this.isPlaying = false; }}为了方便测试，我自己用Java 写了一个UDP的服务器，其功能非常的弱，就是接收，然后转发给另一方：
import java.io.IOException;import java.net.DatagramPacket;import java.net.DatagramSocket;import java.net.InetAddress;import java.net.SocketException;import java.net.UnknownHostException;public class AudioServer implements Runnable { DatagramSocket socket; DatagramPacket packet;// 从客户端接收到的UDP包 DatagramPacket sendPkt;// 转发给另一个客户端的UDP包 byte[] pktBuffer = new byte[1024]; int bufferSize = 1024; boolean isRunning = false; int myport = 5656; // /////////// String clientIpStr = "192.168.1.104"; InetAddress clientIp; int clientPort = 5757; public AudioServer() { try { clientIp = InetAddress.getByName(clientIpStr); } catch (UnknownHostException e1) { e1.printStackTrace(); } try { socket = new DatagramSocket(myport); packet = new DatagramPacket(pktBuffer, bufferSize); } catch (SocketException e) { e.printStackTrace(); } System.out.println("服务器初始化完成"); } public void startServer() { this.isRunning = true; new Thread(this).start(); } public void run() { try { while (isRunning) { socket.receive(packet); sendPkt = new DatagramPacket(packet.getData(), packet.getLength(), packet.getAddress(), clientPort); socket.send(sendPkt); try { Thread.sleep(20); } catch (InterruptedException e) { e.printStackTrace(); } } } catch (IOException e) { } } // main public static void main(String[] args) { new AudioServer().startServer(); }}5.结语：
Android使用 ILBC 进行语音通话的大致过程就讲述完了，此系列只是做一个ILBC 使用原理的介绍，距离真正的语音
通话还有很多工作要做，缺点还是很多的：
1. 文章中介绍的只是单方通话，如果要做成双方互相通话或者一对多的通话，就需要增加更多的流程处理，其服务端
也要做很多工作；
2. 实时性：本程序在局域网中使用时，实时性还是较高的，但是再广域网中，效果可能会有所下降，除此之外，本
程序还缺少时间戳的处理，如果网络状况不理想，或者数据延迟，就会导致语音播放前后混乱；
3. 服务器很弱：真正的流媒体服务器，需要很强的功能，来对数据进行处理，我是为了方便，就写了一个简单的，
最近打算移植live555，用来做专门的流媒体服务器，用RTP协议对数据进行封装，这样效果应该会好很多。