在.NET中使用Speex -- 音訊資料編解碼（轉）

　Speex是一套開源的音訊編解碼庫，最新版本還包含了迴音消除和防抖動等功能，如果我們想開發語音聊天或視訊會議這樣的系統，Speex將是一個不錯的選擇。到 http://www.speex.org可以下載Speex的原始碼（編譯後的dll為libspeex.dll），最新版本為1.2。不過原始碼是用C++開發的，直接在.NET中使用會有諸多不便，為此，我用C#將其封裝，使得編解碼的呼叫相當簡單。

　　由於Speex原始匯出的API不是很方便C#呼叫，所以，在用C#封裝之前，先要用C++對Speex的原始API進行簡化，新建一個名為Speex的VC專案，然後引用libspeex.dll的相關庫檔案，新增cpp檔案後，複製下列原始碼到檔案中：

<!--Code highlighting produced by Actipro CodeHighlighter (freeware)http://www.CodeHighlighter.com/--&gt#include "speex\speex.h"
#include <windows.h>
#include <stdio.h>
#include <stdlib.h>

#include "speex/speex_echo.h"
#include "speex/speex_preprocess.h"
#include "Speex.h"

#define FRAME_SIZE 160

float encoder_input[FRAME_SIZE];
void *encoder_state;
SpeexBits encoder_bits;

BOOL APIENTRY DllMain( HANDLE hModule,
                       DWORD  ul_reason_for_call,
                       LPVOID lpReserved
                     )
{
    return TRUE;
}

extern "C" __declspec(dllexport) void encoder_init(int quality)
{
    encoder_state = speex_encoder_init(&speex_nb_mode);
    speex_encoder_ctl(encoder_state, SPEEX_SET_QUALITY, &quality);
    speex_bits_init(&encoder_bits);
}

extern "C" __declspec(dllexport) void encoder_dispose()
{
    speex_encoder_destroy(encoder_state);
    speex_bits_destroy(&encoder_bits);
}

extern "C" __declspec(dllexport) int encoder_encode(const short *data, char *output)
{
    for (int i = 0; i < FRAME_SIZE; i++)
        encoder_input[i] = data[i];
    speex_bits_reset(&encoder_bits);
    speex_encode(encoder_state, encoder_input, &encoder_bits);
    return speex_bits_write(&encoder_bits, output, 200);
}

float decoder_output[FRAME_SIZE];
void *decoder_state;
SpeexBits decoder_bits;

extern "C" __declspec(dllexport) void decoder_init()
{
    decoder_state = speex_decoder_init(&speex_nb_mode);
    int tmp = 1;
    speex_decoder_ctl(decoder_state, SPEEX_SET_ENH, &tmp);
    speex_bits_init(&decoder_bits);
}
extern "C" __declspec(dllexport) void decoder_dispose()
{
    speex_decoder_destroy(decoder_state);
    speex_bits_destroy(&decoder_bits);
}
extern "C" __declspec(dllexport) void decoder_decode(int nbBytes, char *data, short *output)
{
    speex_bits_read_from(&decoder_bits, data, nbBytes);
    speex_decode(decoder_state, &decoder_bits, decoder_output);
    for (int i = 0; i < FRAME_SIZE; i++)
    {
        output[i] = decoder_output[i];
    }
}

/***************************************************  迴音消除 **************************************/

bool      m_bSpeexEchoHasInit;
SpeexEchoState*   m_SpeexEchoState;
SpeexPreprocessState* m_pPreprocessorState;
int      m_nFilterLen;
int      m_nSampleRate;
float*   m_pfNoise;

extern "C" __declspec(dllexport) void SpeexEchoCapture(short* input_frame, short* output_frame)
{
    speex_echo_capture(m_SpeexEchoState, input_frame, output_frame);
}

extern "C" __declspec(dllexport) void SpeexEchoPlayback(short* echo_frame)
{
    speex_echo_playback(m_SpeexEchoState, echo_frame);
}

extern "C" __declspec(dllexport) void SpeexEchoReset()
{
    if (m_SpeexEchoState != NULL)
    {
        speex_echo_state_destroy(m_SpeexEchoState);
        m_SpeexEchoState = NULL;
    }
    if (m_pPreprocessorState != NULL)
    {
        speex_preprocess_state_destroy(m_pPreprocessorState);
        m_pPreprocessorState = NULL;
    }
    if (m_pfNoise != NULL)
    {
        delete []m_pfNoise;
        m_pfNoise = NULL;
    }
    m_bSpeexEchoHasInit = false;
}

extern "C" __declspec(dllexport) void SpeexEchoInit(int filter_length, int sampling_rate ,bool associatePreprocesser)
{
    SpeexEchoReset();

    if (filter_length<=0 || sampling_rate<=0)
    {
      m_nFilterLen  = 160*8;
      m_nSampleRate = 8000;
    }
    else
    {
      m_nFilterLen  = filter_length;
      m_nSampleRate = sampling_rate;
    }

    m_SpeexEchoState = speex_echo_state_init(FRAME_SIZE, m_nFilterLen);
    m_pPreprocessorState = speex_preprocess_state_init(FRAME_SIZE, m_nSampleRate);
    if(associatePreprocesser)
    {
        speex_preprocess_ctl(m_pPreprocessorState, SPEEX_PREPROCESS_SET_ECHO_STATE,m_SpeexEchoState);
    }
    m_pfNoise = new float[FRAME_SIZE+1];
    m_bSpeexEchoHasInit = true;
}

extern "C" __declspec(dllexport) void SpeexEchoDoAEC(short* mic, short* ref, short* out)
{
    if (!m_bSpeexEchoHasInit)
    {
      return;
    }

    speex_echo_cancellation(m_SpeexEchoState,(const __int16 *) mic,(const __int16 *) ref,(__int16 *) out);
    }

　　編譯便生成Speex.dll。

　　如果對VC不熟悉也沒關係，文末會直接給出libspeex.dll和Speex.dll的下載，直接使用就OK了。

　　現在，C#可以呼叫Speex.dll匯出的簡單函式了，最終封裝的原始碼如下：

<!--Code highlighting produced by Actipro CodeHighlighter (freeware)http://www.CodeHighlighter.com/--&gt ///

    /// 對Speex的C#封裝。
    /// zhuweisky 2010.05.13
    ///
    public class Speex :IAudioCodec
    {
        private const int FrameSize = 160;

        #region IsDisposed
        private volatile bool isDisposed = false;
        public bool IsDisposed
        {
            get { return isDisposed; }
        }
        #endregion

        #region Ctor
        ///

        /// 初始化。
        ///
        /// 編碼質量，取值0~10
        public Speex(int quality)
        {
            if (quality < 0 || quality > 10)
            {
                throw new Exception("quality value must be between 0 and 10.");
            }

            Speex.encoder_init(quality);
            Speex.decoder_init();
        }
        #endregion

        #region Dispose
        public void Dispose()
        {
            this.isDisposed = true;
            System.Threading.Thread.Sleep(100);
            Speex.decoder_dispose();
            Speex.encoder_dispose();
        }
        #endregion

        #region Encode
        ///

        /// 將採集到的音訊資料進行編碼。
        ///
        public byte[] Encode(byte[] data)
        {
            if (this.isDisposed)
            {
                return null;
            }

            if (data.Length % (FrameSize * 2) != 0)
            {
                throw new ArgumentException("Invalid Data Length.");
            }

            int nbBytes;
            short[] input = new short[FrameSize];
            byte[] buffer = new byte[200];
            byte[] output = new byte[0];
            for (int i = 0; i < data.Length / (FrameSize * 2); i++)
            {
                for (int j = 0; j < input.Length; j++)
                {
                    input[j] = (short)(data[i * FrameSize * 2 + j * 2] + data[i * FrameSize * 2 + j * 2 + 1] * 0x100);
                }

                nbBytes = Speex.encoder_encode(input, buffer);
                Array.Resize<byte>(ref output, output.Length + nbBytes + sizeof(int));
                Array.Copy(buffer, 0, output, output.Length - nbBytes, nbBytes);

                for (int j = 0; j < sizeof(int); j++)
                {
                    output[output.Length - nbBytes - sizeof(int) + j] = (byte)(nbBytes % 0x100);
                    nbBytes /= 0x100;
                }
            }
            return output;
        }
        #endregion

        #region Decode
        ///

        /// 將編碼後的資料進行解碼得到原始的音訊資料。
        ///
        public byte[] Decode(byte[] data)
        {
            if (this.isDisposed)
            {
                return null;
            }

            int nbBytes, index = 0;
            byte[] input;
            short[] buffer = new short[FrameSize];
            byte[] output = new byte[0];
            while (index < data.Length)
            {
                nbBytes = 0;
                index += sizeof(int);
                for (int i = 1; i <= sizeof(int); i++)
                    nbBytes = nbBytes * 0x100 + data[index - i];
                input = new byte[nbBytes];
                Array.Copy(data, index, input, 0, input.Length);
                index += input.Length;
                Speex.decoder_decode(nbBytes, input, buffer);
                Array.Resize<byte>(ref output, output.Length + FrameSize * 2);
                for (int i = 0; i < FrameSize; i++)
                {
                    output[output.Length - FrameSize * 2 + i * 2] = (byte)(buffer[i] % 0x100);
                    output[output.Length - FrameSize * 2 + i * 2 + 1] = (byte)(buffer[i] / 0x100);
                }
            }
            return output;
        }
        #endregion

        #region Pinvoke
        [DllImport("Speex.dll", EntryPoint = "encoder_init")]
        internal extern static void encoder_init(int quality);
        [DllImport("Speex.dll", EntryPoint = "encoder_dispose")]
        internal extern static void encoder_dispose();
        [DllImport("Speex.dll", EntryPoint = "encoder_encode")]
        internal extern static int encoder_encode(short[] data, byte[] output);
        [DllImport("Speex.dll", EntryPoint = "decoder_init")]
        internal extern static void decoder_init();
        [DllImport("Speex.dll", EntryPoint = "decoder_dispose")]
        internal extern static void decoder_dispose();
        [DllImport("Speex.dll", EntryPoint = "decoder_decode")]
        internal extern static void decoder_decode(int nbBytes, byte[] data, short[] output);
        #endregion
    }

只有四個方法：Initialize、Encode、Decode、Dispose。方法引數的含義也非常明顯。

　一般音訊對話的整個流程是這樣的：採集 -> 編碼 -> 網路傳輸 -> 解碼 -> 播放。

　而該封裝的Speex類解決了這個過程中的音訊編碼和解碼的問題。你可以複製該原始碼到你的專案，並將從http://www.speex.org下載的speex.dll放到執行目錄下，就可以正常地使用SPEEX的編解碼功能了。

　關於Speex更高階的功能，我正在研究中，有興趣的朋友可以email給我一起探討。　　

　　下載 libspeex.dll和Speex.dll。

來自 “ ITPUB部落格 ” ，連結：http://blog.itpub.net/12639172/viewspace-674149/，如需轉載，請註明出處，否則將追究法律責任。

在.NET中使用Speex -- 音訊資料編解碼 （轉）

相關文章

在.NET中使用Speex -- 音訊資料編解碼（轉）