iOS音訊程式設計之實時語音通訊（對講機功能）

weixin_34208283發表於2018-10-15
iOS音訊程式設計

需求：手機通過Mic採集PCM編碼的原始音訊資料，將PCM轉換為AAC編碼格式，通過MultipeerConnectivity框架連線手機併傳送AAC資料，在接收端使用Audio Queue播放收到的AAC音訊。

技術點

AAC Converter, Audio Unit, MultipeerConnectivity

音訊設定

對音訊以44.1KHZ的取樣率來取樣,以64000的位元率對PCM進行AAC轉碼

1）對AVAudioSession的設定

NSError *error;
self.session = [AVAudioSession sharedInstance];
[self.session setCategory:AVAudioSessionCategoryPlayAndRecord error:&error];
handleError(error);
//route變化監聽
[[NSNotificationCenter defaultCenter] addObserver:self selector:@selector(audioSessionRouteChangeHandle:) name:AVAudioSessionRouteChangeNotification object:self.session];

[self.session setPreferredIOBufferDuration:0.005 error:&error];
handleError(error);
[self.session setPreferredSampleRate:kSmaple error:&error];
handleError(error);

//[self.session overrideOutputAudioPort:AVAudioSessionPortOverrideSpeaker error:&error];
//handleError(error);

[self.session setActive:YES error:&error];
handleError(error);

-(void)audioSessionRouteChangeHandle:(NSNotification *)noti{
//    NSError *error;
//    [self.session overrideOutputAudioPort:AVAudioSessionPortOverrideSpeaker error:&error];
//    handleError(error);
[self.session setActive:YES error:nil];
if (self.startRecord) {
    CheckError(AudioOutputUnitStart(_toneUnit), "couldnt start audio unit");
    }
}

音訊輸入輸出路徑改變會觸發audioSessionRouteChangeHandle,如果想一直讓音訊從手機的揚聲器輸出需要在每次Route改變時，把音訊輸出重定向到AVAudioSessionPortOverrideSpeaker,否則為手機聽筒輸出音訊;其他設定說明請參照iOS音訊程式設計之變聲處理的初始化部分

2)對Audio Unit的設定

AudioComponentDescription acd;
acd.componentType = kAudioUnitType_Output;
acd.componentSubType = kAudioUnitSubType_RemoteIO;
acd.componentFlags = 0;
acd.componentFlagsMask = 0;
acd.componentManufacturer = kAudioUnitManufacturer_Apple;
AudioComponent inputComponent = AudioComponentFindNext(NULL, &acd);
AudioComponentInstanceNew(inputComponent, &_toneUnit);


UInt32 enable = 1;
AudioUnitSetProperty(_toneUnit,
                     kAudioOutputUnitProperty_EnableIO,
                     kAudioUnitScope_Input,
                     kInputBus,
                     &enable,
                     sizeof(enable));


mAudioFormat.mSampleRate         = kSmaple;//取樣率
mAudioFormat.mFormatID           = kAudioFormatLinearPCM;//PCM取樣
mAudioFormat.mFormatFlags        = kAudioFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked;
mAudioFormat.mFramesPerPacket    = 1;//每個資料包多少幀
mAudioFormat.mChannelsPerFrame   = 1;//1單聲道，2立體聲
mAudioFormat.mBitsPerChannel     = 16;//語音每取樣點佔用位數
mAudioFormat.mBytesPerFrame      = mAudioFormat.mBitsPerChannel*mAudioFormat.mChannelsPerFrame/8;//每幀的bytes數
mAudioFormat.mBytesPerPacket     = mAudioFormat.mBytesPerFrame*mAudioFormat.mFramesPerPacket;//每個資料包的bytes總數，每幀的bytes數＊每個資料包的幀數
mAudioFormat.mReserved           = 0;

CheckError(AudioUnitSetProperty(_toneUnit,
                                kAudioUnitProperty_StreamFormat,
                                kAudioUnitScope_Output, kInputBus,
                                &mAudioFormat, sizeof(mAudioFormat)),
           "couldn't set the remote I/O unit's input client format");

CheckError(AudioUnitSetProperty(_toneUnit,
                                kAudioOutputUnitProperty_SetInputCallback,
                                kAudioUnitScope_Output,
                                kInputBus,
                                &_inputProc, sizeof(_inputProc)),
           "couldnt set remote i/o render callback for input");


CheckError(AudioUnitInitialize(_toneUnit),
           "couldn't initialize the remote I/O unit");

具體引數說明請參照iOS音訊程式設計之變聲處理

採集音訊資料的輸入回撥

static OSStatus inputRenderTone(
                     void *inRefCon,
                     AudioUnitRenderActionFlags     *ioActionFlags,
                     const AudioTimeStamp         *inTimeStamp,
                     UInt32                         inBusNumber,
                     UInt32                         inNumberFrames,
                     AudioBufferList             *ioData)

{

VoiceConvertHandle *THIS=(__bridge VoiceConvertHandle*)inRefCon;

AudioBufferList bufferList;
bufferList.mNumberBuffers = 1;
bufferList.mBuffers[0].mData = NULL;
bufferList.mBuffers[0].mDataByteSize = 0;
OSStatus status = AudioUnitRender(THIS->_toneUnit,
                                  ioActionFlags,
                                  inTimeStamp,
                                  kInputBus,
                                  inNumberFrames,
                                  &bufferList);

NSInteger lastTimeRear = recordStruct.rear;
for (int i = 0; i < inNumberFrames; i++) {
    SInt16 data = ((SInt16 *)bufferList.mBuffers[0].mData)[i];
    recordStruct.recordArr[recordStruct.rear] = data;
    recordStruct.rear = (recordStruct.rear+1)%kRecordDataLen;
    }
if ((lastTimeRear/1024 + 1) == (recordStruct.rear/1024)) {
     pthread_cond_signal(&recordCond);
    }
return status;
}

採用迴圈佇列儲存原始的音訊資料，每1024點的PCM資料，讓Converter轉換為AAC編碼,所以當收集了1024點PCM後，喚醒Converter執行緒。

3)音訊轉碼

初始化

AudioStreamBasicDescription sourceDes = mAudioFormat;
AudioStreamBasicDescription targetDes;
memset(&targetDes, 0, sizeof(targetDes));
targetDes.mFormatID = kAudioFormatMPEG4AAC;
targetDes.mSampleRate = kSmaple;
targetDes.mChannelsPerFrame = sourceDes.mChannelsPerFrame;
UInt32 size = sizeof(targetDes);
CheckError(AudioFormatGetProperty(kAudioFormatProperty_FormatInfo,
                                  0, NULL, &size, &targetDes),
           "couldnt create target data format");


//選擇軟體編碼
AudioClassDescription audioClassDes;
CheckError(AudioFormatGetPropertyInfo(kAudioFormatProperty_Encoders,
                                      sizeof(targetDes.mFormatID),
                                      &targetDes.mFormatID,
                                      &size), "cant get kAudioFormatProperty_Encoders");
UInt32 numEncoders = size/sizeof(AudioClassDescription);
AudioClassDescription audioClassArr[numEncoders];
CheckError(AudioFormatGetProperty(kAudioFormatProperty_Encoders,
                                  sizeof(targetDes.mFormatID),
                                  &targetDes.mFormatID,
                                  &size,
                                  audioClassArr),
           "wrirte audioClassArr fail");
for (int i = 0; i < numEncoders; i++) {
    if (audioClassArr[i].mSubType == kAudioFormatMPEG4AAC
        && audioClassArr[i].mManufacturer == kAppleSoftwareAudioCodecManufacturer) {
        memcpy(&audioClassDes, &audioClassArr[i], sizeof(AudioClassDescription));
        break;
    }
}

CheckError(AudioConverterNewSpecific(&sourceDes, &targetDes, 1,
                                     &audioClassDes, &_encodeConvertRef),
           "cant new convertRef");

size = sizeof(sourceDes);
CheckError(AudioConverterGetProperty(_encodeConvertRef, kAudioConverterCurrentInputStreamDescription, &size, &sourceDes), "cant get kAudioConverterCurrentInputStreamDescription");

size = sizeof(targetDes);
CheckError(AudioConverterGetProperty(_encodeConvertRef, kAudioConverterCurrentOutputStreamDescription, &size, &targetDes), "cant get kAudioConverterCurrentOutputStreamDescription");

UInt32 bitRate = 64000;
size = sizeof(bitRate);
CheckError(AudioConverterSetProperty(_encodeConvertRef,
                                     kAudioConverterEncodeBitRate,
                                     size, &bitRate),
           "cant set covert property bit rate");
[self performSelectorInBackground:@selector(convertPCMToAAC) withObject:nil];

主要是設定編碼器的輸入音訊格式(PCM),輸出音訊格式(AAC),選擇軟體編碼器(預設使用硬體編碼器),設定編碼器的位元率

AAC編碼

-(void)convertPCMToAAC{
UInt32 maxPacketSize = 0;
UInt32 size = sizeof(maxPacketSize);
CheckError(AudioConverterGetProperty(_encodeConvertRef,
                                     kAudioConverterPropertyMaximumOutputPacketSize,
                                     &size,
                                     &maxPacketSize),
           "cant get max size of packet");

AudioBufferList *bufferList = malloc(sizeof(AudioBufferList));
bufferList->mNumberBuffers = 1;
bufferList->mBuffers[0].mNumberChannels = 1;
bufferList->mBuffers[0].mData = malloc(maxPacketSize);
bufferList->mBuffers[0].mDataByteSize = maxPacketSize;

for (; ; ) {
    @autoreleasepool {


    pthread_mutex_lock(&recordLock);
    while (ABS(recordStruct.rear - recordStruct.front) < 1024 ) {
        pthread_cond_wait(&recordCond, &recordLock);
    }
    pthread_mutex_unlock(&recordLock);

    SInt16 *readyData = (SInt16 *)calloc(1024, sizeof(SInt16));
    memcpy(readyData, &recordStruct.recordArr[recordStruct.front], 1024*sizeof(SInt16));
    recordStruct.front = (recordStruct.front+1024)%kRecordDataLen;
    UInt32 packetSize = 1;
    AudioStreamPacketDescription *outputPacketDescriptions = malloc(sizeof(AudioStreamPacketDescription)*packetSize);
    bufferList->mBuffers[0].mDataByteSize = maxPacketSize;
    CheckError(AudioConverterFillComplexBuffer(_encodeConvertRef,
                                               encodeConverterComplexInputDataProc,
                                               readyData,
                                               &packetSize,
                                               bufferList,
                                               outputPacketDescriptions),
               "cant set AudioConverterFillComplexBuffer");
    free(outputPacketDescriptions);
    free(readyData);

    NSMutableData *fullData = [NSMutableData dataWithBytes:bufferList->mBuffers[0].mData length:bufferList->mBuffers[0].mDataByteSize];

    if ([self.delegate respondsToSelector:@selector(covertedData:)]) {
        [self.delegate covertedData:[fullData copy]];
    }
    }
}

新建的bufferList是用來存放每次轉碼後的AAC音訊資料.for迴圈中等待音訊輸入回撥存滿1024個PCM陣列並喚醒它。outputPacketDescriptions陣列是每次轉換的AAC編碼後各個包的描述,但這裡每次只轉換一包資料(由傳入的packetSize決定)。呼叫AudioConverterFillComplexBuffer觸發轉碼，他的第二個引數是填充原始音訊資料的回撥。轉碼完成後，會將轉碼的資料存放在它的第五個引數中(bufferList).轉換完成的AAC就可以傳送給另外一臺手機了。

填充原始資料回撥

OSStatus encodeConverterComplexInputDataProc(AudioConverterRef inAudioConverter,
                                         UInt32 *ioNumberDataPackets,
                                         AudioBufferList *ioData,
                                         AudioStreamPacketDescription **outDataPacketDescription,
                                         void *inUserData)
{
    ioData->mBuffers[0].mData = inUserData;
    ioData->mBuffers[0].mNumberChannels = 1;
    ioData->mBuffers[0].mDataByteSize = 1024*2;
       *ioNumberDataPackets = 1024;
    return 0;
}

4）Audio Queue播放AAC音訊資料

Audio Queue基礎知識

音訊資料以一個個AudioQueueBuffer的形式存在與音訊佇列中，Audio Queue使用它提供的音訊資料來播放，某一個AudioQueueBuffer使用完畢後，會呼叫Audio Queue的回撥，要求使用者再在這個AudioQueueBuffer填入資料，並使它加入Audio Queue中，如此迴圈，達到不間斷播放音訊資料的效果。

Audio Queue初始化

CheckError(AudioQueueNewOutput(&targetDes,
                               fillBufCallback,
                               (__bridge void *)self,
                               NULL,
                               NULL,
                               0,
                               &(_playQueue)),
           "cant new audio queue");
CheckError( AudioQueueSetParameter(_playQueue,
                                   kAudioQueueParam_Volume, 1.0),
           "cant set audio queue gain");

for (int i = 0; i < 3; i++) {
    AudioQueueBufferRef buffer;
    CheckError(AudioQueueAllocateBuffer(_playQueue, 1024, &buffer), "cant alloc buff");
    BNRAudioQueueBuffer *buffObj = [[BNRAudioQueueBuffer alloc] init];
    buffObj.buffer = buffer;
    [_buffers addObject:buffObj];
    [_reusableBuffers addObject:buffObj];
}
[self performSelectorInBackground:@selector(playData) withObject:nil];

Audio Queue播放音訊資料

-(void)playData{
    for (; ; ) {
    @autoreleasepool {

    NSMutableData *data = [[NSMutableData alloc] init];
    pthread_mutex_lock(&playLock);
    if (self.aacArry.count%8 != 0 || self.aacArry.count == 0) {
        pthread_cond_wait(&playCond, &playLock);
    }
    AudioStreamPacketDescription *paks = calloc(sizeof(AudioStreamPacketDescription), 8);
    for (int i = 0; i < 8 ; i++) {//8包AAC資料組成放入一個AudioQueueBuffer的資料包
        BNRAudioData *audio = [self.aacArry firstObject];
        [data appendData:audio.data];
        paks[i].mStartOffset = audio.packetDescription.mStartOffset;
        paks[i].mDataByteSize = audio.packetDescription.mDataByteSize;
        [self.aacArry removeObjectAtIndex:0];
    }
    pthread_mutex_unlock(&playLock);

    pthread_mutex_lock(&buffLock);
    if (_reusableBuffers.count == 0) {
        static dispatch_once_t onceToken;
        dispatch_once(&onceToken, ^{
            AudioQueueStart(_playQueue, nil);
        });
        pthread_cond_wait(&buffcond, &buffLock);

    }
    BNRAudioQueueBuffer *bufferObj = [_reusableBuffers firstObject];
    [_reusableBuffers removeObject:bufferObj];
    pthread_mutex_unlock(&buffLock);

    memcpy(bufferObj.buffer->mAudioData,[data bytes] , [data length]);
    bufferObj.buffer->mAudioDataByteSize = (UInt32)[data length];
    CheckError(AudioQueueEnqueueBuffer(_playQueue, bufferObj.buffer, 8, paks), "cant enqueue");
    free(paks);

    }
    }
}

static void fillBufCallback(void *inUserData,
                       AudioQueueRef inAQ,
                       AudioQueueBufferRef buffer){
VoiceConvertHandle *THIS=(__bridge VoiceConvertHandle*)inUserData;

for (int i = 0; i < THIS->_buffers.count; ++i) {
    if (buffer == [THIS->_buffers[i] buffer]) {
        pthread_mutex_lock(&buffLock);
        [THIS->_reusableBuffers addObject:THIS->_buffers[i]];
        pthread_mutex_unlock(&buffLock);
        pthread_cond_signal(&buffcond);
        break;
    }
    }   
}

在playData中等待收到的aacArry資料，這裡要注意:每1024點PCM轉換成的一包AAC資料加入到AudioQueueBuffer中，不足以使Audio Queue播放音訊，所以這裡使用8包AAC資料放到一個AudioQueueBuffer中。fillBufCallback是Audio Queue播放完一個AudioQueueBuffer呼叫的回撥函式，在這裡面通知playData可以往使用完的AudioQueueBufferRef填資料了，填完後，用AudioQueueEnqueueBuffer將它加入Audio Queue中，這個三個AudioQueueBufferRef不斷重用。

實時語音通訊處理

原來是想用藍芽來傳送資料的，但是自己寫的藍芽傳送資料機制的速度跟不上轉換的AAC資料。使用MultipeerConnectivity框架既可使用藍芽也可以使用WIFI來通訊，底層自動選擇。當把兩個手機的WIFI都關掉時，他們使用藍芽來傳送資料，在剛剛建立通話時，能聽到傳送的語音，之後就聽不到了，使用wifi傳輸資料時不會出現這種情況。

  1. MultipeerConnectivity基礎知識

MCNearbyServiceAdvertiser傳送廣播，並接收MCNearbyServiceBrowser端的邀請,MCSession傳送接收資料、管理連線狀態。建立連線和通訊的流程是，MCNearbyServiceAdvertiser廣播服務，MCNearbyServiceBrowser搜到這個服務後，要求把這個服務所對用的MCPeerID加入到它自己(MCNearbyServiceBrowser端)的MCSession中，MCNearbyServiceAdvertiser收到這個邀請，並同意，同時也將MCNearbyServiceBrowser端對應的MCPeerID加入到了它自己(MCNearbyServiceAdvertiser)的MCSession中.
之後雙方可以使用各自的MCSession傳送接收資料。

2）各端傳送本身轉碼的AAC資料，並接收對方傳送的AAC資料提供給Auduio queue播放

原始碼下載地址

相關文章