一、语音聊天说专业点就是即时语音,是一种基于网络的快速传递语音信息的技术,普遍应用于各类社交软件中,优势主要有以下几点:
(1)时效性:视频直播会因为带宽问题有时出现延迟高的问题,而语音直播相对来说会好很多,延迟低,并且能够第·一时间与听众互动,时效性强。
(2)隐私性:这一点体现在何处,如主播不想暴露自己的长相,或者进行问题回答是,没有视频的话会让主播感到更安心,所以语音直播隐私性更强。
(3)内容质量高:因为语音直播不靠“颜值”只有好的内容才能够吸引用户,所以语音直播相对来说内容质量更高。
(4)成本降低:语音直播相对视频直播来说,带宽流量等都会便宜许多,成本降低不少,更加实惠。
二、语音聊天主要步骤:音频采集、压缩编码、网络传输、解码还原、播放音频,如下图所示
下面就从代码的角度来详说一下这几个步骤。
(1)音频采集,读取麦克风设备数据
1 private readonly WaveIn _waveIn; 2 _waveIn = new WaveIn(); 3 _waveIn.BufferMilliseconds = 50; 4 _waveIn.DeviceNumber = 0; 5 _waveIn.DataAvailable += OnAudioCaptured; 6 _waveIn.StartRecording();
(2)音频数据压缩编码,常见压缩格式比较多,例如mp3、acc、speex等,这里以speex为例
1 private readonly WideBandSpeexCodec _speexCodec; 2 _speexCodec = new WideBandSpeexCodec(); 3 _waveIn.WaveFormat = _speexCodec.RecordFormat; 4 5 void OnAudioCaptured(object sender, WaveInEventArgs e) 6 { 7 byte[] encoded = _speexCodec.Encode(e.Buffer, 0, e.BytesRecorded); 8 _audioClient.Send(encoded); 9 }
(3)网络传输,为了保证即时传输udp协议有着天然的优点
1 using SAEA.Sockets; 2 using SAEA.Sockets.Base; 3 using SAEA.Sockets.Model; 4 using System; 5 using System.Net; 6 7 namespace GFF.Component.GAudio.Net 8 { 9 public class AudioClient 10 { 11 IClientSocket _udpClient; 12 13 BaseUnpacker _baseUnpacker; 14 15 public event Action<Byte[]> OnReceive; 16 17 public AudioClient(IPEndPoint endPoint) 18 { 19 var bContext = new BaseContext(); 20 21 _udpClient = SocketFactory.CreateClientSocket(SocketOptionBuilder.Instance.SetSocket(SAEASocketType.Udp) 22 .SetIPEndPoint(endPoint) 23 .UseIocp(bContext) 24 .SetReadBufferSize(SocketOption.UDPMaxLength) 25 .SetWriteBufferSize(SocketOption.UDPMaxLength) 26 .Build()); 27 28 _baseUnpacker = (BaseUnpacker)bContext.Unpacker; 29 30 _udpClient.OnReceive += _udpClient_OnReceive; 31 } 32 33 private void _udpClient_OnReceive(byte[] data) 34 { 35 OnReceive?.Invoke(data); 36 } 37 38 public void Connect() 39 { 40 _udpClient.Connect(); 41 } 42 43 public void Send(byte[] data) 44 { 45 _udpClient.SendAsync(data); 46 } 47 48 public void Disconnect() 49 { 50 _udpClient.Disconnect(); 51 } 52 53 } 54 }
(4)服务器转发,客户端使用udp,服务器这里同样也使用udp来转发
1 using SAEA.Sockets; 2 using SAEA.Sockets.Base; 3 using SAEA.Sockets.Interface; 4 using SAEA.Sockets.Model; 5 using System; 6 using System.Collections.Concurrent; 7 using System.Net; 8 using System.Threading.Tasks; 9 10 namespace GFF.Component.GAudio.Net 11 { 12 public class AudioServer 13 { 14 IServerSocket _udpServer; 15 16 ConcurrentDictionary<string, IUserToken> _cache; 17 18 public AudioServer(IPEndPoint endPoint) 19 { 20 _cache = new ConcurrentDictionary<string, IUserToken>(); 21 22 _udpServer = SocketFactory.CreateServerSocket(SocketOptionBuilder.Instance.SetSocket(SAEASocketType.Udp) 23 .SetIPEndPoint(endPoint) 24 .UseIocp<BaseContext>() 25 .SetReadBufferSize(SocketOption.UDPMaxLength) 26 .SetWriteBufferSize(SocketOption.UDPMaxLength) 27 .SetTimeOut(5000) 28 .Build()); 29 _udpServer.OnAccepted += _udpServer_OnAccepted; 30 _udpServer.OnDisconnected += _udpServer_OnDisconnected; 31 _udpServer.OnReceive += _udpServer_OnReceive; 32 } 33 34 public void Start() 35 { 36 _udpServer.Start(); 37 } 38 39 public void Stop() 40 { 41 _udpServer.Stop(); 42 } 43 44 private void _udpServer_OnReceive(ISession currentSession, byte[] data) 45 { 46 Parallel.ForEach(_cache.Keys, (id) => 47 { 48 try 49 { 50 _udpServer.SendAsync(id, data); 51 } 52 catch { } 53 }); 54 } 55 56 private void _udpServer_OnAccepted(object obj) 57 { 58 var ut = (IUserToken)obj; 59 if (ut != null) 60 { 61 _cache.TryAdd(ut.ID, ut); 62 } 63 } 64 65 private void _udpServer_OnDisconnected(string ID, Exception ex) 66 { 67 _cache.TryRemove(ID, out IUserToken _); 68 } 69 } 70 }
(5)解码还原,客户端将从服务器收到的数据按约定的压缩格式,进行解压缩还原成音频数据
1 private readonly BufferedWaveProvider _waveProvider; 2 _waveProvider = new BufferedWaveProvider(_speexCodec.RecordFormat); 3 4 private void _audioClient_OnReceive(byte[] data) 5 { 6 byte[] decoded = _speexCodec.Decode(data, 0, data.Length); 7 _waveProvider.AddSamples(decoded, 0, decoded.Length); 8 }
(6)播放音频,使用播放设备来播放解码后的音频数据
1 private readonly IWavePlayer _waveOut; 2 _waveOut = new WaveOut(); 3 _waveOut.Init(_waveProvider); 4 _waveOut.Play();
三、测试运行,通过分析语音聊天的几个关键问题点后,按步骤封装好代码,接下来就是用实例来测试一下效果了。
客户端封装在按钮事件中:
1 GAudioClient _gAudioClient = null; 2 3 private void toolStripDropDownButton2_ButtonClick(object sender, EventArgs e) 4 { 5 if (_gAudioClient == null) 6 { 7 ClientConfig clientConfig = ClientConfig.Instance(); 8 _gAudioClient = new GAudioClient(clientConfig.IP, clientConfig.Port + 2); 9 _gAudioClient.Start(); 10 } 11 else 12 { 13 _gAudioClient.Dispose(); 14 _gAudioClient = null; 15 } 16 }
服务端封装在main函数中:
1 ConsoleHelper.WriteLine("正在初始化语音服务器...", ConsoleColor.DarkBlue); 2 _gAudioServer = new GAudioServer(filePort + 1); 3 ConsoleHelper.WriteLine("语音服务器初始化完毕...", ConsoleColor.DarkBlue); 4 ConsoleHelper.WriteLine("正在启动语音服务器...", ConsoleColor.DarkBlue); 5 _gAudioServer.Start(); 6 ConsoleHelper.WriteLine("语音服务器初始化完毕", ConsoleColor.DarkBlue);
万事俱备,现在F5跑起来试试。
如上红框所示,喊了几句相当于Hello World的Hello没有问题,大功初步告成~
转载请标明本文来源:https://www.cnblogs.com/yswenli/p/14353482.html
更多内容欢迎我的的github:https://github.com/yswenli/GFF
如果发现本文有什么问题和任何建议,也随时欢迎交流~