AAC 到 PCM 音頻解碼

時間 2019-11-11

標籤 aac pcm 音頻解碼简体版

原文原文鏈接

最近遇到在 iOS 平臺上實時播放 AAC 音頻數據流, 一開始嘗試用 AudioQueue 直接解 AAC 未果, 轉而將 AAC 解碼爲 PCM, 最終實現了 AAC 實時流在 iOS 平臺下的播放問題.ios

AAC 轉 PCM 須要藉助解碼庫來實現, 目前瞭解到有兩個庫能幹這個事 : faad 和 ffmpeg.git

faad 算是輕量級的解碼庫, 編譯出來全平臺靜態庫文件大小 2M 左右, API 也比較簡單, 缺點是功能單一隻處理 AAC , 它還有一個對應的編碼庫叫 faac.
ffmpeg 體積龐大, 功能豐富, API 略顯複雜.

下面分別梳理使用這兩個庫完成解碼的過程.github

###faad

下載源碼

#下載
wget http://downloads.sourceforge.net/faac/faad2-2.7.tar.gz
#解壓縮
tar xvzf faad2-2.7.tar.gz
#重命名
mv faad2-2.7 faad

寫編譯腳本, vi build-faad.sh

#!/bin/sh

CONFIGURE_FLAGS="--enable-static --with-pic"

ARCHS="arm64 armv7s armv7 x86_64 i386"

# directories
SOURCE="faad"
FAT="fat-faad"

SCRATCH="scratch-faad"
# must be an absolute path
THIN=`pwd`/"thin-faad"

COMPILE="y"
LIPO="y"

if [ "$*" ]
then
if [ "$*" = "lipo" ]
then
# skip compile
COMPILE=
else
ARCHS="$*"
if [ $# -eq 1 ]
then
# skip lipo
LIPO=
fi
fi
fi

if [ "$COMPILE" ]
then
CWD=`pwd`
for ARCH in $ARCHS
do
echo "building $ARCH..."
mkdir -p "$SCRATCH/$ARCH"
cd "$SCRATCH/$ARCH"

if [ "$ARCH" = "i386" -o "$ARCH" = "x86_64" ]
then
PLATFORM="iPhoneSimulator"
CPU=
if [ "$ARCH" = "x86_64" ]
then
SIMULATOR="-mios-simulator-version-min=7.0"
HOST=
else
SIMULATOR="-mios-simulator-version-min=5.0"
HOST="--host=i386-apple-darwin"
fi
else
PLATFORM="iPhoneOS"
if [ $ARCH = "armv7s" ]
then
CPU="--cpu=swift"
else
CPU=
fi
SIMULATOR=
HOST="--host=arm-apple-darwin"
fi

XCRUN_SDK=`echo $PLATFORM | tr '[:upper:]' '[:lower:]'`
CC="xcrun -sdk $XCRUN_SDK clang -Wno-error=unused-command-line-argument-hard-error-in-future"
AS="$CWD/$SOURCE/extras/gas-preprocessor.pl $CC"
CFLAGS="-arch $ARCH $SIMULATOR"
CXXFLAGS="$CFLAGS"
LDFLAGS="$CFLAGS"

CC=$CC CFLAGS=$CXXFLAGS LDFLAGS=$LDFLAGS CPPFLAGS=$CXXFLAGS CXX=$CC CXXFLAGS=$CXXFLAGS  $CWD/$SOURCE/configure \
$CONFIGURE_FLAGS \
$HOST \
--prefix="$THIN/$ARCH" \
--disable-shared \
--without-mp4v2

make clean && make && make install-strip
cd $CWD
done
fi

if [ "$LIPO" ]
then
echo "building fat binaries..."
mkdir -p $FAT/lib
set - $ARCHS
CWD=`pwd`
cd $THIN/$1/lib
for LIB in *.a
do
cd $CWD
lipo -create `find $THIN -name $LIB` -output $FAT/lib/$LIB
done

cd $CWD
cp -rf $THIN/$1/include $FAT
fi

保存編譯腳本到解壓出的 faad 目錄同一級目錄下, 並添加可執行權限 chmod a+x build-faad.shobjective-c

編譯 ./build-faad.sh 當前目錄下 fat-faad 即爲編譯結果所在位置, 裏面有頭文件和支持全平臺(armv7, armv7s ,i386, x86_64, arm64)的靜態庫shell
添加靜態庫到工程依賴 (鼠標拖 fat-faad 目錄到 xcode 工程目錄下), 建立解碼文件FAACDecoder.h,FAACDecoder.mswift
FAACDecoder.hxcode

//
//  FAACDecoder.h
//  EasyClient
//
//  Created by 吳鵬 on 16/9/3.
//  Copyright © 2016年 EasyDarwin. All rights reserved.
//

#ifndef FAACDecoder_h
#define FAACDecoder_h

typedef struct {
    NeAACDecHandle handle;
    int sample_rate;
    int channels;
    int bit_rate;
}FAADContext;

FAADContext* faad_decoder_create(int sample_rate, int channels, int bit_rate);
int faad_decode_frame(FAADContext *pParam, unsigned char *pData, int nLen, unsigned char *pPCM, unsigned int *outLen);
void faad_decode_close(FAADContext *pParam);

#endif /* FAACDecoder_h */

FAACDecoder.m

//
//  FAACDecoder.m
//  EasyClient
//
//  Created by 吳鵬 on 16/9/3.
//  Copyright © 2016年 EasyDarwin. All rights reserved.
//
#import <Foundation/Foundation.h>
#import "FAACDecoder.h"
#import "faad.h"

uint32_t _get_frame_length(const unsigned char *aac_header)
{
    uint32_t len = *(uint32_t *)(aac_header + 3);
    len = ntohl(len); //Little Endian
    len = len << 6;
    len = len >> 19;
    return len;
}

FAADContext* faad_decoder_create(int sample_rate, int channels, int bit_rate)
{
    NeAACDecHandle handle = NeAACDecOpen();
    if(!handle){
        printf("NeAACDecOpen failed\n");
        goto error;
    }
    NeAACDecConfigurationPtr conf = NeAACDecGetCurrentConfiguration(handle);
    if(!conf){
        printf("NeAACDecGetCurrentConfiguration failed\n");
        goto error;
    }
    conf->defSampleRate = sample_rate;
    conf->outputFormat = FAAD_FMT_16BIT;
    conf->dontUpSampleImplicitSBR = 1;
    NeAACDecSetConfiguration(handle, conf);
    
    FAADContext* ctx = malloc(sizeof(FAADContext));
    ctx->handle = handle;
    ctx->sample_rate = sample_rate;
    ctx->channels = channels;
    ctx->bit_rate = bit_rate;
    return ctx;
    
error:
    if(handle){
        NeAACDecClose(handle);
    }
    return NULL;
}

int faad_decode_frame(FAADContext *p, unsigned char *pData, int nLen, unsigned char *pPCM, unsigned int *outLen)
{
    FAADContext* pCtx = (FAADContext*)pParam;
    NeAACDecHandle handle = pCtx->handle;
    long res = NeAACDecInit(handle, pData, nLen, (unsigned long*)&pCtx->sample_rate, (unsigned char*)&pCtx->channels);
    if (res < 0) {
        printf("NeAACDecInit failed\n");
        return -1;
    }
    NeAACDecFrameInfo info;
    uint32_t framelen = _get_frame_length(pData);
    unsigned char *buf = (unsigned char *)NeAACDecDecode(handle, &info, pData, framelen);
    if (buf && info.error == 0) {
        if (info.samplerate == 44100) {
            //src: 2048 samples, 4096 bytes
            //dst: 2048 samples, 4096 bytes
            int tmplen = (int)info.samples * 16 / 8;
            memcpy(pPCM,buf,tmplen);
            *outLen = tmplen;
        } else if (info.samplerate == 22050) {
            //src: 1024 samples, 2048 bytes
            //dst: 2048 samples, 4096 bytes
            short *ori = (short*)buf;
            short tmpbuf[info.samples * 2];
            int tmplen = (int)info.samples * 16 / 8 * 2;
            for (int32_t i = 0, j = 0; i < info.samples; i += 2) {
                tmpbuf[j++] = ori[i];
                tmpbuf[j++] = ori[i + 1];
                tmpbuf[j++] = ori[i];
                tmpbuf[j++] = ori[i + 1];
            }
            memcpy(pPCM,tmpbuf,tmplen);
            *outLen = tmplen;
        }else if(info.samplerate == 8000){
            //從雙聲道的數據中提取單通道
            for(int i=0,j=0; i<4096 && j<2048; i+=4, j+=2)
            {
                pPCM[j]= buf[i];
                pPCM[j+1]=buf[i+1];
            }
            *outLen = (unsigned int)info.samples;
        }
    } else {
        printf("NeAACDecDecode failed\n");
        return -1;
    }
    return 0;
}

void faad_decode_close(void *pParam)
{
    if(!pParam){
        return;
    }
    FAADContext* pCtx = (FAADContext*)pParam;
    if(pCtx->handle){
        NeAACDecClose(pCtx->handle);
    }
    free(pCtx);
}

幾個主要 API :app

NeAACDecOpen
NeAACDecGetCurrentConfiguration
NeAACDecSetConfiguration
NeAACDecInit
NeAACDecDecode
NeAACDecClose

###ffmpeg

下載編譯參考 https://github.com/kewlbear/FFmpeg-iOS-build-scriptui
添加 ffmpeg 靜態庫到工程依賴, 建立解碼文件AACDecoder.h, AACDecoder.m編碼
AACDecoder.h

#ifndef _AACDecoder_h
#define _AACDecoder_h

void *aac_decoder_create(int sample_rate, int channels, int bit_rate);
int aac_decode_frame(void *pParam, unsigned char *pData, int nLen, unsigned char *pPCM, unsigned int *outLen);
void aac_decode_close(void *pParam);

#endif

AACDecoder.m

#include "AACDecoder.h"
#include "libavformat/avformat.h"
#include "libswresample/swresample.h"
#include "libavcodec/avcodec.h"

typedef struct AACDFFmpeg {
    AVCodecContext *pCodecCtx;
    AVFrame *pFrame;
    struct SwrContext *au_convert_ctx;
    int out_buffer_size;
} AACDFFmpeg;

void *aac_decoder_create(int sample_rate, int channels, int bit_rate)
{
    AACDFFmpeg *pComponent = (AACDFFmpeg *)malloc(sizeof(AACDFFmpeg));
    AVCodec *pCodec = avcodec_find_decoder(AV_CODEC_ID_AAC);
    if (pCodec == NULL)
    {
        printf("find aac decoder error\r\n");
        return 0;
    }
    // 建立顯示contedxt
    pComponent->pCodecCtx = avcodec_alloc_context3(pCodec);
    pComponent->pCodecCtx->channels = channels;
    pComponent->pCodecCtx->sample_rate = sample_rate;
    pComponent->pCodecCtx->bit_rate = bit_rate;
    if(avcodec_open2(pComponent->pCodecCtx, pCodec, NULL) < 0)
    {
        printf("open codec error\r\n");
        return 0;
    }
    
    pComponent->pFrame = av_frame_alloc();
    

    uint64_t out_channel_layout = channels < 2 ? AV_CH_LAYOUT_MONO:AV_CH_LAYOUT_STEREO;
    int out_nb_samples = 1024;
    enum AVSampleFormat out_sample_fmt = AV_SAMPLE_FMT_S16;
    
    pComponent->au_convert_ctx = swr_alloc();
    pComponent->au_convert_ctx = swr_alloc_set_opts(pComponent->au_convert_ctx, out_channel_layout, out_sample_fmt, sample_rate,
                                      out_channel_layout, AV_SAMPLE_FMT_FLTP, sample_rate, 0, NULL);
    swr_init(pComponent->au_convert_ctx);
    int out_channels = av_get_channel_layout_nb_channels(out_channel_layout);
    pComponent->out_buffer_size = av_samples_get_buffer_size(NULL, out_channels, out_nb_samples, out_sample_fmt, 1);

    return (void *)pComponent;
}

int aac_decode_frame(void *pParam, unsigned char *pData, int nLen, unsigned char *pPCM, unsigned int *outLen)
{
    AACDFFmpeg *pAACD = (AACDFFmpeg *)pParam;
    AVPacket packet;
    av_init_packet(&packet);
    
    packet.size = nLen;
    packet.data = pData;
    
    int got_frame = 0;
    int nRet = 0;
    if (packet.size > 0)
    {
        nRet = avcodec_decode_audio4(pAACD->pCodecCtx, pAACD->pFrame, &got_frame, &packet);
        if (nRet < 0)
        {
   printf("avcodec_decode_audio4:%d\r\n",nRet);
            printf("avcodec_decode_audio4 %d  sameles = %d  outSize = %d\r\n", nRet, pAACD->pFrame->nb_samples, pAACD->out_buffer_size);
            return nRet;
        }

        if(got_frame)
        {
            swr_convert(pAACD->au_convert_ctx, &pPCM, pAACD->out_buffer_size, (const uint8_t **)pAACD->pFrame->data, pAACD->pFrame->nb_samples);
            *outLen = pAACD->out_buffer_size;
        }
    }

    av_free_packet(&packet);
    if (nRet > 0)
    {
        return 0;
    }
    return -1;
}

void aac_decode_close(void *pParam)
{
    AACDFFmpeg *pComponent = (AACDFFmpeg *)pParam;
    if (pComponent == NULL)
    {
        return;
    }
    
    swr_free(&pComponent->au_convert_ctx);
    
    if (pComponent->pFrame != NULL)
    {
        av_frame_free(&pComponent->pFrame);
        pComponent->pFrame = NULL;
    }
    
    if (pComponent->pCodecCtx != NULL)
    {
        avcodec_close(pComponent->pCodecCtx);
        avcodec_free_context(&pComponent->pCodecCtx);
        pComponent->pCodecCtx = NULL;
    }
    
    free(pComponent);
}

相關標籤/搜索

每日一句

每一个你不满意的现在，都有一个你没有努力的曾经。