public/rtphone
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

- clang format + naming improved

Browse Source
This commit is contained in:
Dmytro Bogovych
2026-09-03 10:03:42 +03:00
parent 249e614def
commit e166bae1ea
204 changed files with 20621 additions and 21019 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/engine/media/MT_AudioStream.cpp
+74 -48
View File
@@ -18,13 +18,12 @@
#define LOG_SUBSYSTEM "media"
//#define DUMP_SENDING_AUDIO
// #define DUMP_SENDING_AUDIO
using namespace MT;
AudioStream::AudioStream(const CodecList::Settings& settings)
:mPacketTime(0), mEncodedTime(0), mCodecSettings(settings),
mRemoteTelephoneCodec(0), mRtpSession(), mTransmittingPayloadType(-1),
mRtpSender(mStat)
: mPacketTime(0), mEncodedTime(0), mCodecSettings(settings), mRemoteTelephoneCodec(0), mRtpSession(),
mTransmittingPayloadType(-1), mRtpSender(mStat)
{
mOutputBuffer.setCapacity(16384);
mCapturedAudio.setCapacity(16384);
@@ -36,19 +35,19 @@ AudioStream::AudioStream(const CodecList::Settings& settings)
// Configure transmitter
jrtplib::RTPExternalTransmissionParams params(&mRtpSender, 0);
jrtplib::RTPSessionParams sessionParams;
jrtplib::RTPSessionParams sessionParams;
sessionParams.SetAcceptOwnPackets(true);
sessionParams.SetMaximumPacketSize(MT_MAXRTPPACKET);
sessionParams.SetResolveLocalHostname(false);
sessionParams.SetUsePollThread(false);
sessionParams.SetOwnTimestampUnit(1/8000.0);
sessionParams.SetOwnTimestampUnit(1 / 8000.0);
mRtpSession.Create(sessionParams, &params, jrtplib::RTPTransmitter::ExternalProto);
mRtpDtmfSession.Create(sessionParams, &params, jrtplib::RTPTransmitter::ExternalProto);
// Attach srtp session to sender
mRtpSender.setSrtpSession(&mSrtpSession);
//mRtpDump = new RtpDump("d:\\outgoing.rtp");
//mRtpSender.setDumpWriter(mRtpDump);
// mRtpDump = new RtpDump("d:\\outgoing.rtp");
// mRtpSender.setDumpWriter(mRtpDump);
#if defined(DUMP_SENDING_AUDIO)
mSendingDump = std::make_shared<WavFileWriter>();
@@ -107,7 +106,7 @@ void AudioStream::setDestination(const RtpPair<InternetAddress>& dest)
mRtpSender.setDestination(dest);
}
void AudioStream::setTransmittingCodec(Codec::Factory& factory, int payloadType)
void AudioStream::setTransmittingCodec(Codec::Factory& factory, int payloadType)
{
ICELogInfo(<< "Selected codec " << factory.name() << "/" << factory.samplerate() << " for transmitting");
@@ -146,28 +145,38 @@ void AudioStream::addData(const void* buffer, int bytes)
{
Lock l(mMutex);
codec = mTransmittingCodec.get();
if (nullptr == codec) {
if (nullptr == codec)
{
// ICELogDebug(<< "No transmitting codec selected.");
return;
}
}
// Resample
unsigned dstlen = unsigned(float(codec->samplerate() / float(AUDIO_SAMPLERATE)) * bytes);
unsigned dstlen = unsigned(float(codec->samplerate() / float(AUDIO_SAMPLERATE)) * bytes);
Audio::Resampler* r = nullptr;
switch (codec->samplerate())
{
case 8000: r = &mCaptureResampler8; break;
case 16000: r = &mCaptureResampler16; break;
case 32000: r = &mCaptureResampler32; break;
case 48000: r = &mCaptureResampler48; break;
case 8000:
r = &mCaptureResampler8;
break;
case 16000:
r = &mCaptureResampler16;
break;
case 32000:
r = &mCaptureResampler32;
break;
case 48000:
r = &mCaptureResampler48;
break;
default:
assert(0);
}
size_t processedInput = 0;
dstlen = r->processBuffer(buffer, bytes, processedInput, mResampleBuffer, dstlen);
// ProcessedInput output value is ignored - because sample rate of input is always 8/16/32/48K - so all buffer is processed
// ProcessedInput output value is ignored - because sample rate of input is always 8/16/32/48K - so all buffer is
// processed
// See if we need stereo <-> mono conversions
unsigned stereolen = 0;
@@ -181,22 +190,26 @@ void AudioStream::addData(const void* buffer, int bytes)
// See if inband dtmf audio should be sent instead
ByteBuffer dtmf;
if (mDtmfContext.type() == DtmfContext::Dtmf_Inband && mDtmfContext.getInband(AUDIO_MIC_BUFFER_LENGTH, codec->samplerate(), dtmf))
if (mDtmfContext.type() == DtmfContext::Dtmf_Inband &&
mDtmfContext.getInband(AUDIO_MIC_BUFFER_LENGTH, codec->samplerate(), dtmf))
mCapturedAudio.add(dtmf.data(), dtmf.size());
else
mCapturedAudio.add(stereolen ? mStereoBuffer : mResampleBuffer, stereolen ? stereolen : dstlen);
// See if it is time to send RFC2833 tone
ByteBuffer rfc2833, stopPacket;
if (mDtmfContext.type() == DtmfContext::Dtmf_Rfc2833 && mDtmfContext.getRfc2833(AUDIO_MIC_BUFFER_LENGTH, rfc2833, stopPacket))
if (mDtmfContext.type() == DtmfContext::Dtmf_Rfc2833 &&
mDtmfContext.getRfc2833(AUDIO_MIC_BUFFER_LENGTH, rfc2833, stopPacket))
{
if (rfc2833.size())
mRtpDtmfSession.SendPacket(rfc2833.data(), rfc2833.size(), mRemoteTelephoneCodec, true, AUDIO_MIC_BUFFER_LENGTH * 8);
mRtpDtmfSession.SendPacket(rfc2833.data(), rfc2833.size(), mRemoteTelephoneCodec, true,
AUDIO_MIC_BUFFER_LENGTH * 8);
if (stopPacket.size())
{
for (int i=0; i<3; i++)
mRtpDtmfSession.SendPacket(stopPacket.data(), stopPacket.size(), mRemoteTelephoneCodec, true, AUDIO_MIC_BUFFER_LENGTH * 8);
for (int i = 0; i < 3; i++)
mRtpDtmfSession.SendPacket(stopPacket.data(), stopPacket.size(), mRemoteTelephoneCodec, true,
AUDIO_MIC_BUFFER_LENGTH * 8);
}
}
@@ -205,14 +218,15 @@ void AudioStream::addData(const void* buffer, int bytes)
int packetTime = mPacketTime ? mPacketTime : codec->frameTime();
// Make stereo version if required
for (int i=0; i<mCapturedAudio.filled() / codec->pcmLength(); i++)
for (int i = 0; i < mCapturedAudio.filled() / codec->pcmLength(); i++)
{
if (mSendingDump)
mSendingDump->write((const char*)mCapturedAudio.data() + codec->pcmLength() * i, codec->pcmLength());
auto r = codec->encode({(const uint8_t*)mCapturedAudio.data() + codec->pcmLength()*i, (size_t)codec->pcmLength()},
{(uint8_t*)mFrameBuffer, MT_MAXAUDIOFRAME});
auto r =
codec->encode({(const uint8_t*)mCapturedAudio.data() + codec->pcmLength() * i, (size_t)codec->pcmLength()},
{(uint8_t*)mFrameBuffer, MT_MAXAUDIOFRAME});
// Counter of processed input bytes of raw pcm data from microphone
processed += codec->pcmLength();
encodedTime += codec->frameTime();
@@ -224,9 +238,10 @@ void AudioStream::addData(const void* buffer, int bytes)
if (packetTime <= encodedTime)
{
// Time to send packet
ICELogMedia(<< "Sending RTP packet pt = " << mTransmittingPayloadType << ", plength = " << (int)mEncodedAudio.size() << " to ");
ICELogMedia(<< "Sending RTP packet pt = " << mTransmittingPayloadType
<< ", plength = " << (int)mEncodedAudio.size() << " to ");
mRtpSession.SendPacketEx(mEncodedAudio.data(), mEncodedAudio.size(), mTransmittingPayloadType, false,
packetTime * codec->samplerate()/1000, 0, nullptr, 0);
packetTime * codec->samplerate() / 1000, 0, nullptr, 0);
mEncodedAudio.clear();
encodedTime = 0;
}
@@ -239,14 +254,14 @@ void AudioStream::addData(const void* buffer, int bytes)
void AudioStream::copyDataTo(Audio::Mixer& mixer, int needed)
{
// mStreamMap is also mutated from the network thread (dataArrived)
Lock l(mMutex);
Lock l(mMutex);
// Local audio mixer - used to send audio to media observer
Audio::Mixer localMixer;
Audio::Mixer localMixer;
Audio::DataWindow forObserver;
// Iterate
for (auto& streamIter: mStreamMap)
for (auto& streamIter : mStreamMap)
{
Audio::DataWindow w;
w.setCapacity(32768);
@@ -287,7 +302,8 @@ void AudioStream::copyDataTo(Audio::Mixer& mixer, int needed)
{
int mixedBytes = localMixer.mixAndGetPcm(forObserver);
if (mixedBytes > 0)
mMediaObserver->onMedia(forObserver.data(), mixedBytes, MT::Stream::MediaDirection::Incoming, this, mMediaObserverTag);
mMediaObserver->onMedia(forObserver.data(), mixedBytes, MT::Stream::MediaDirection::Incoming, this,
mMediaObserverTag);
}
}
@@ -295,11 +311,12 @@ void AudioStream::dataArrived(PDatagramSocket s, const void* buffer, int length,
{
// Protects mStreamMap (also iterated by copyDataTo on the audio thread)
// and the receive/decrypt buffers.
Lock l(mMutex);
Lock l(mMutex);
jrtplib::RTPIPv6Address addr6;
jrtplib::RTPIPv4Address addr4;
jrtplib::RTPExternalTransmissionInfo* info = dynamic_cast<jrtplib::RTPExternalTransmissionInfo*>(mRtpSession.GetTransmissionInfo());
jrtplib::RTPIPv6Address addr6;
jrtplib::RTPIPv4Address addr4;
jrtplib::RTPExternalTransmissionInfo* info =
dynamic_cast<jrtplib::RTPExternalTransmissionInfo*>(mRtpSession.GetTransmissionInfo());
assert(info);
// Drop RTP packets if stream is not receiving now; let RTCP go
@@ -315,15 +332,16 @@ void AudioStream::dataArrived(PDatagramSocket s, const void* buffer, int length,
if (mSrtpSession.active())
{
bool srtpResult;
size_t srcLength = length; size_t dstLength = sizeof mSrtpDecodeBuffer;
bool srtpResult;
size_t srcLength = length;
size_t dstLength = sizeof mSrtpDecodeBuffer;
if (RtpHelper::isRtp(mReceiveBuffer, receiveLength))
srtpResult = mSrtpSession.unprotectRtp(mReceiveBuffer, srcLength, mSrtpDecodeBuffer, &dstLength);
else
srtpResult = mSrtpSession.unprotectRtcp(mReceiveBuffer, srcLength, mSrtpDecodeBuffer, &dstLength);
if (!srtpResult)
{
ICELogError(<<"Cannot decrypt SRTP packet.");
ICELogError(<< "Cannot decrypt SRTP packet.");
return;
}
@@ -379,8 +397,9 @@ void AudioStream::dataArrived(PDatagramSocket s, const void* buffer, int length,
// Find right handler for rtp stream
SingleAudioStream* rtpStream = nullptr;
auto streamIter = mStreamMap.find(packet->GetSSRC());
if (streamIter == mStreamMap.end()) {
auto streamIter = mStreamMap.find(packet->GetSSRC());
if (streamIter == mStreamMap.end())
{
rtpStream = new SingleAudioStream(mCodecSettings, mStat);
mStreamMap.insert({packet->GetSSRC(), rtpStream});
}
@@ -395,14 +414,13 @@ void AudioStream::dataArrived(PDatagramSocket s, const void* buffer, int length,
// without these, a skewed or buggy peer can poison mRttDelay
// (and therefore the Id term in MOS).
double rtt = mRtpSession.GetCurrentSourceInfo()->INF_GetRoundtripTime().GetDouble();
if (rtt > 0 && rtt < 30.0) // reject "RTT not making any sense" (>30s)
if (rtt > 0 && rtt < 30.0) // reject "RTT not making any sense" (>30s)
{
// Once an average is established, cap a new sample at 3x mean
// so a single outlier can't skew the running RTT.
constexpr double kRttNormalizeFactor = 3.0;
const double meanRtt = mStat.mRttDelay.average();
if (mStat.mRttDelay.is_initialized() && meanRtt > 0.0 &&
rtt > meanRtt * kRttNormalizeFactor)
const double meanRtt = mStat.mRttDelay.average();
if (mStat.mRttDelay.is_initialized() && meanRtt > 0.0 && rtt > meanRtt * kRttNormalizeFactor)
{
rtt = meanRtt * kRttNormalizeFactor;
}
@@ -438,8 +456,12 @@ void AudioStream::readFile(const Audio::PWavFileReader& stream, MediaDirection d
{
switch (direction)
{
case MediaDirection::Outgoing: mDumpStreams.mStreamForReadingOutgoing = stream; break;
case MediaDirection::Incoming: mDumpStreams.mStreamForReadingIncoming = stream; break;
case MediaDirection::Outgoing:
mDumpStreams.mStreamForReadingOutgoing = stream;
break;
case MediaDirection::Incoming:
mDumpStreams.mStreamForReadingIncoming = stream;
break;
}
}
@@ -447,8 +469,12 @@ void AudioStream::writeFile(const Audio::PWavFileWriter& writer, MediaDirection
{
switch (direction)
{
case MediaDirection::Outgoing: mDumpStreams.mStreamForRecordingOutgoing = writer; break;
case MediaDirection::Incoming: mDumpStreams.mStreamForRecordingIncoming = writer; break;
case MediaDirection::Outgoing:
mDumpStreams.mStreamForRecordingOutgoing = writer;
break;
case MediaDirection::Incoming:
mDumpStreams.mStreamForRecordingIncoming = writer;
break;
}
}