Skip to content

Commit

Permalink
mp3 config decode and bug fix (#358)
Browse files Browse the repository at this point in the history 
* mp3 config decode and bug fix

* revert audio transrater changes

* review comments

* revert ts fix in audiodecoder worker

* return error message
  • Loading branch information
@yezhan10
yezhan10 authored Oct 16, 2024
1 parent 1bd336c commit b4e640b
Show file tree
Hide file tree
Showing 7 changed files with 243 additions and 15 deletions.
1 change: 1 addition & 0 deletions CMakeLists.txt
View file
Original file line number Diff line number Diff line change
Expand Up @@ -206,6 +206,7 @@ add_executable(MediaServerUnitTest
${CMAKE_CURRENT_LIST_DIR}/test/unit/TestAAC.cpp
${CMAKE_CURRENT_LIST_DIR}/test/unit/TestMP3.cpp
${CMAKE_CURRENT_LIST_DIR}/test/unit/TestOpus.cpp
${CMAKE_CURRENT_LIST_DIR}/test/unit/TestMP3Config.cpp
${CMAKE_CURRENT_LIST_DIR}/test/unit/TestAudioPipe.cpp
${CMAKE_CURRENT_LIST_DIR}/test/unit/TestAMFNumber.cpp
${CMAKE_CURRENT_LIST_DIR}/test/unit/TestVideoLayersAllocation.cpp
Expand Down
2 changes: 1 addition & 1 deletion include/AudioBuffer.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -85,7 +85,7 @@ class AudioBuffer
uint16_t numSamplesPerFrame;
uint8_t numChannels;
uint64_t ts = 0;
uint16_t clockRate = 0;
uint32_t clockRate = 0;
std::vector<int16_t> pcmBuffer;
};

Expand Down
9 changes: 3 additions & 6 deletions src/aac/AACDecoder.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -63,14 +63,15 @@ AACDecoder::~AACDecoder()

bool AACDecoder::SetConfig(const uint8_t* data,const size_t size)
{
if (inited) return true;
AACSpecificConfig config;

Debug("-AACDecoder::SetConfig()\n");

//Decode it
if (!config.Decode(data,size))
//Error
return false;
return Error("AACDecoder::SetConfig() Could not parse AAC config\n");;

//Set data
ctx->channels = config.GetChannels();
Expand All @@ -83,7 +84,6 @@ bool AACDecoder::SetConfig(const uint8_t* data,const size_t size)
memcpy(side,data,size);
//We are inited
inited = true;

//Done
return true;
}
Expand All @@ -93,10 +93,7 @@ int AACDecoder::Decode(const AudioFrame::const_shared& audioFrame)
//Check we have config
if (audioFrame->HasCodecConfig())
SetConfig(audioFrame->GetCodecConfigData(), audioFrame->GetCodecConfigSize());

if (!inited)
return Error("-AACDecoder::Decode() Not inited\n");


//Set data
packet->data = audioFrame ? (uint8_t*)audioFrame->GetData() : nullptr;
packet->size = audioFrame ? audioFrame->GetLength() : 0;
Expand Down
133 changes: 133 additions & 0 deletions src/mp3/MP3Config.h
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,133 @@
#ifndef MP3CONFIG_H
#define MP3CONFIG_H

#include "config.h"
#include "bitstream/BitReader.h"
#include "bitstream/BitWriter.h"

class MP3Config
{
public:
enum class MPEGAudioVersion {
MPEGVersion2,
MPEGVersion1,
MPEGVersionReserved
};
enum class MPEGAudioLayer {
MPEGLayer3,
MPEGLayerReserved
};

static constexpr std::array<DWORD, 3> Mpeg1SamplingRates = {44100, 48000, 32000};
static constexpr std::array<DWORD, 15> Mpeg1Bitrates = {0, 32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320};

static constexpr std::array<DWORD, 3> Mpeg2SamplingRates = {22050, 24000, 16000};
static constexpr std::array<DWORD, 15> Mpeg2Bitrates = {0, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160};
public:

MP3Config() = default;
MP3Config(DWORD rate, BYTE channels)
{
this->samplingRate = rate;
this->channels = channels;
}

bool Decode(const BYTE* data,const DWORD size)
{
if (size != 4)
return Error("-MP3Config::Decode() config data must be size of 4 bytes\n");

//Create bit reader
BufferReader reader(data,size);
BitReader r(reader);

try{
auto frameSync = r.Get(11);
if (frameSync != 0x7FF)
return Error("-MP3Config::Decode() Wrong FrameSync\n");

auto ver = r.Get(2);
switch (ver) {
case 2:
audioVersion = MPEGAudioVersion::MPEGVersion2;
break;
case 3:
audioVersion = MPEGAudioVersion::MPEGVersion1;
break;
default:
return Error("-MP3Config::Decode() MPEG audio version not supported\n");
};
auto layer = r.Get(2);
if (layer == 1)
{
audioLayer = MPEGAudioLayer::MPEGLayer3;
frameLength = audioVersion == MPEGAudioVersion::MPEGVersion1 ? 1152 : 576;
}
else
{
return Error("-MP3Config::Decode() MPEG audio layer not supported\n");
}
r.Skip(1);

auto bitrateIdx = r.Get(4);
if (bitrateIdx >= 15)
return Error("-MP3Config::Decode() bitrate idx out of range\n");
else if (bitrateIdx == 0)
return Error("-MP3Config::Decode() free bitrate idx not supported\n");

bitrate = audioVersion == MPEGAudioVersion::MPEGVersion1 ? Mpeg1Bitrates[bitrateIdx] : Mpeg2Bitrates[bitrateIdx];

auto samplingRateIdx = r.Get(2);
if (samplingRateIdx >= 3)
return Error("-MP3Config::Decode() invalid sampling rate idx\n");

samplingRate = audioVersion == MPEGAudioVersion::MPEGVersion1 ? Mpeg1SamplingRates[samplingRateIdx] : Mpeg2SamplingRates[samplingRateIdx];

padding = r.Get(1);
if (padding)
paddingSize = 1;

r.Skip(1);
auto channelMode = r.Get(2);
if (channelMode == 3)
channels = 1;
else
channels = 2;
}
catch (std::exception& e)
{
return false;
}
return true;
}

void Dump() const
{
Debug("[MP3Config \n");
Debug("\t mpegVersion=%u\n" , audioVersion);
Debug("\t mpegLayer=%u\n" , audioLayer);
Debug("\t rate=%u\n" , samplingRate);
Debug("\t channels=%u\n" , channels);
Debug("\t paddingSize=%u\n" , paddingSize);
Debug("\t frameLength=%u\n" , frameLength);
Debug("/]\n");
}
MPEGAudioVersion GetAudioVersion() const {return audioVersion;}
MPEGAudioLayer GetAudioLayer() const {return audioLayer;}
DWORD GetRate() const { return samplingRate; }
BYTE GetChannels() const { return channels; }
DWORD GetPaddingSize() const { return padding ? paddingSize:0; }
DWORD GetFrameLength() const { return frameLength; }
DWORD GetBitrate() const { return bitrate * 1000; }

private:
MPEGAudioVersion audioVersion = MPEGAudioVersion::MPEGVersionReserved;
MPEGAudioLayer audioLayer = MPEGAudioLayer::MPEGLayerReserved;
bool padding;
BYTE paddingSize;
DWORD samplingRate = 0;
DWORD bitrate = 0;
BYTE channels = 0;
DWORD frameLength = 0;
};
#endif /* MPEGCONFIG_H */
20 changes: 12 additions & 8 deletions src/mp3/MP3Decoder.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@ extern "C" {
}
#include "log.h"
#include "MP3Decoder.h"

#include "MP3Config.h"


MP3Decoder::MP3Decoder()
Expand Down Expand Up @@ -34,9 +34,6 @@ MP3Decoder::MP3Decoder()
packet = av_packet_alloc();
//Allocate frame
frame = av_frame_alloc();

//Get the number of samples
numFrameSamples = 1024;
}

MP3Decoder::~MP3Decoder()
Expand All @@ -59,14 +56,24 @@ MP3Decoder::~MP3Decoder()

bool MP3Decoder::SetConfig(const uint8_t* data,const size_t size)
{
if (inited) return true;
MP3Config config;


Debug("-MP3Decoder::SetConfig()\n");

//Decode it
if (!config.Decode(data, size))
//Error
return Error("MP3Decoder::SetConfig() Could not parse MP3 config\n");;
//Set side data pon packet
uint8_t *side = av_packet_new_side_data(packet, AV_PKT_DATA_NEW_EXTRADATA, size);
//Copy it
memcpy(side,data,size);
//We are inited
inited = true;

ctx->channels = config.GetChannels();
ctx->sample_rate = config.GetRate();

//Done
return true;
Expand All @@ -78,9 +85,6 @@ int MP3Decoder::Decode(const AudioFrame::const_shared& audioFrame)
//Check we have config
if (audioFrame->HasCodecConfig())
SetConfig(audioFrame->GetCodecConfigData(), audioFrame->GetCodecConfigSize());

if (!inited)
return Error("-MP3Decoder::Decode() Not inited\n");

//Set data
packet->data = audioFrame ? (uint8_t*)audioFrame->GetData() : nullptr;
Expand Down
1 change: 1 addition & 0 deletions test/unit/TestAudioPipe.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -35,6 +35,7 @@ void helperTestAudioPipe(const AudioPipeParam& playParam, const AudioPipeParam&
auto audioBuffer = std::make_shared<AudioBuffer>(playFrameSize, numChannels);
audioBuffer->SetSamples(inLoc, playFrameSize*numChannels);
audioBuffer->SetTimestamp(playPTS[i]);
audioBuffer->SetClockRate(playSampleRate);
audPipe.PlayBuffer(audioBuffer);
inLoc += playFrameSize*numChannels;
};
Expand Down
92 changes: 92 additions & 0 deletions test/unit/TestMP3Config.cpp
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,92 @@
#include "TestCommon.h"

#include "config.h"
#include "log.h"
#include "mp3/MP3Config.h"
#include <algorithm>

TEST(TestMP3Config, SupportedMP3Config)
{
{
BYTE mp3[4] = { 0xff, 0xfb, 0x74, 0xc4 };
MP3Config config;
ASSERT_TRUE(config.Decode(mp3, sizeof(mp3)));
/*
[MP3Config
mpegVersion = 1
mpegLayer = layerIII
rate = 48000
channels = 1
paddingSize = 0
frameLength = 1152
/ ]
*/
ASSERT_EQ(config.GetAudioVersion() , MP3Config::MPEGAudioVersion::MPEGVersion1);
ASSERT_EQ(config.GetAudioLayer() , MP3Config::MPEGAudioLayer::MPEGLayer3);
ASSERT_EQ(config.GetRate() , 48000);
ASSERT_EQ(config.GetChannels() , 1);
ASSERT_EQ(config.GetFrameLength() , 1152);
ASSERT_EQ(config.GetBitrate() , 96000);

}
{
BYTE mp3[4] = { 0xff, 0xf3, 0x44, 0xc4 };
MP3Config config;
ASSERT_TRUE(config.Decode(mp3, sizeof(mp3)));
/*
[MP3Config
mpegVersion = 2
mpegLayer = layerIII
rate = 24000
channels = 1
paddingSize = 0
frameLength = 576
bitrate = 96000
/ ]
*/
ASSERT_EQ(config.GetAudioVersion() , MP3Config::MPEGAudioVersion::MPEGVersion2);
ASSERT_EQ(config.GetAudioLayer() , MP3Config::MPEGAudioLayer::MPEGLayer3);
ASSERT_EQ(config.GetRate() , 24000);
ASSERT_EQ(config.GetChannels() , 1);
ASSERT_EQ(config.GetFrameLength() , 576);
ASSERT_EQ(config.GetBitrate() , 32000);
}
{
BYTE mp3[4] = { 0xff, 0xfb, 0x70, 0x64 };
MP3Config config;
ASSERT_TRUE(config.Decode(mp3, sizeof(mp3)));
/*
[MP3Config
mpegVersion = 1
mpegLayer = layerIII
rate = 44100
channels = 2
paddingSize = 0
frameLength = 1152
bitrate = 96000
/ ]
*/
ASSERT_EQ(config.GetAudioVersion() , MP3Config::MPEGAudioVersion::MPEGVersion1);
ASSERT_EQ(config.GetAudioLayer() , MP3Config::MPEGAudioLayer::MPEGLayer3);
ASSERT_EQ(config.GetRate() , 44100);
ASSERT_EQ(config.GetChannels() , 2);
ASSERT_EQ(config.GetFrameLength() , 1152);
ASSERT_EQ(config.GetBitrate() , 96000);
}
}

TEST(TestMP3Config, UnsupportedMP3Config)
{
{
// mpeg version : mpeg2extension
BYTE mp3[4] = { 0xff, 0xe3, 0x74, 0xc4 };
MP3Config config;
ASSERT_FALSE(config.Decode(mp3, sizeof(mp3)));
}
{
// mpeg layer I
BYTE mp3[4] = { 0xff, 0xf7, 0x44, 0xc4 };
MP3Config config;
ASSERT_FALSE(config.Decode(mp3, sizeof(mp3)));
}
}

0 comments on commit b4e640b

Please sign in to comment.