mirror of
https://github.com/danog/libtgvoip.git
synced 2024-11-27 04:34:42 +01:00
a6e6ed5825
Also make opus in-band FEC less aggressive
271 lines
8.3 KiB
C++
Executable File
271 lines
8.3 KiB
C++
Executable File
//
|
|
// libtgvoip is free and unencumbered public domain software.
|
|
// For more information, see http://unlicense.org or the UNLICENSE file
|
|
// you should have received with this source code distribution.
|
|
//
|
|
|
|
#include "OpusEncoder.h"
|
|
#include <assert.h>
|
|
#include <algorithm>
|
|
#include "logging.h"
|
|
#include "VoIPServerConfig.h"
|
|
#ifdef HAVE_CONFIG_H
|
|
#include <opus/opus.h>
|
|
#else
|
|
#include "opus.h"
|
|
#endif
|
|
|
|
namespace{
|
|
int serverConfigValueToBandwidth(int config){
|
|
switch(config){
|
|
case 0:
|
|
return OPUS_BANDWIDTH_NARROWBAND;
|
|
case 1:
|
|
return OPUS_BANDWIDTH_MEDIUMBAND;
|
|
case 2:
|
|
return OPUS_BANDWIDTH_WIDEBAND;
|
|
case 3:
|
|
return OPUS_BANDWIDTH_SUPERWIDEBAND;
|
|
case 4:
|
|
default:
|
|
return OPUS_BANDWIDTH_FULLBAND;
|
|
}
|
|
}
|
|
}
|
|
|
|
tgvoip::OpusEncoder::OpusEncoder(MediaStreamItf *source, bool needSecondary):queue(11), bufferPool(960*2, 10){
|
|
this->source=source;
|
|
source->SetCallback(tgvoip::OpusEncoder::Callback, this);
|
|
enc=opus_encoder_create(48000, 1, OPUS_APPLICATION_VOIP, NULL);
|
|
opus_encoder_ctl(enc, OPUS_SET_COMPLEXITY(10));
|
|
opus_encoder_ctl(enc, OPUS_SET_PACKET_LOSS_PERC(1));
|
|
opus_encoder_ctl(enc, OPUS_SET_INBAND_FEC(1));
|
|
opus_encoder_ctl(enc, OPUS_SET_SIGNAL(OPUS_SIGNAL_VOICE));
|
|
opus_encoder_ctl(enc, OPUS_SET_BANDWIDTH(OPUS_BANDWIDTH_FULLBAND));
|
|
requestedBitrate=20000;
|
|
currentBitrate=0;
|
|
running=false;
|
|
echoCanceller=NULL;
|
|
complexity=10;
|
|
frameDuration=20;
|
|
levelMeter=NULL;
|
|
vadNoVoiceBitrate=static_cast<uint32_t>(ServerConfig::GetSharedInstance()->GetInt("audio_vad_no_voice_bitrate", 6000));
|
|
vadModeVoiceBandwidth=serverConfigValueToBandwidth(ServerConfig::GetSharedInstance()->GetInt("audio_vad_bandwidth", 3));
|
|
vadModeNoVoiceBandwidth=serverConfigValueToBandwidth(ServerConfig::GetSharedInstance()->GetInt("audio_vad_no_voice_bandwidth", 0));
|
|
secondaryEnabledBandwidth=serverConfigValueToBandwidth(ServerConfig::GetSharedInstance()->GetInt("audio_extra_ec_bandwidth", 2));
|
|
secondaryEncoderEnabled=false;
|
|
|
|
if(needSecondary){
|
|
secondaryEncoder=opus_encoder_create(48000, 1, OPUS_APPLICATION_VOIP, NULL);
|
|
opus_encoder_ctl(secondaryEncoder, OPUS_SET_COMPLEXITY(10));
|
|
opus_encoder_ctl(secondaryEncoder, OPUS_SET_SIGNAL(OPUS_SIGNAL_VOICE));
|
|
//opus_encoder_ctl(secondaryEncoder, OPUS_SET_VBR(0));
|
|
opus_encoder_ctl(secondaryEncoder, OPUS_SET_BITRATE(8000));
|
|
opus_encoder_ctl(secondaryEncoder, OPUS_SET_BANDWIDTH(secondaryEnabledBandwidth));
|
|
}else{
|
|
secondaryEncoder=NULL;
|
|
}
|
|
}
|
|
|
|
tgvoip::OpusEncoder::~OpusEncoder(){
|
|
opus_encoder_destroy(enc);
|
|
if(secondaryEncoder)
|
|
opus_encoder_destroy(secondaryEncoder);
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::Start(){
|
|
if(running)
|
|
return;
|
|
running=true;
|
|
thread=new Thread(std::bind(&tgvoip::OpusEncoder::RunThread, this));
|
|
thread->SetName("OpusEncoder");
|
|
thread->Start();
|
|
thread->SetMaxPriority();
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::Stop(){
|
|
if(!running)
|
|
return;
|
|
running=false;
|
|
queue.Put(NULL);
|
|
thread->Join();
|
|
delete thread;
|
|
}
|
|
|
|
|
|
void tgvoip::OpusEncoder::SetBitrate(uint32_t bitrate){
|
|
requestedBitrate=bitrate;
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::Encode(int16_t* data, size_t len){
|
|
if(requestedBitrate!=currentBitrate){
|
|
opus_encoder_ctl(enc, OPUS_SET_BITRATE(requestedBitrate));
|
|
currentBitrate=requestedBitrate;
|
|
LOGV("opus_encoder: setting bitrate to %u", currentBitrate);
|
|
}
|
|
if(levelMeter)
|
|
levelMeter->Update(data, len);
|
|
if(secondaryEncoderEnabled!=wasSecondaryEncoderEnabled){
|
|
wasSecondaryEncoderEnabled=secondaryEncoderEnabled;
|
|
opus_encoder_ctl(enc, OPUS_SET_BANDWIDTH(secondaryEncoderEnabled ? secondaryEnabledBandwidth : OPUS_BANDWIDTH_FULLBAND));
|
|
}
|
|
int32_t r=opus_encode(enc, data, static_cast<int>(len), buffer, 4096);
|
|
if(r<=0){
|
|
LOGE("Error encoding: %d", r);
|
|
}else if(r==1){
|
|
LOGW("DTX");
|
|
}else if(running){
|
|
//LOGV("Packet size = %d", r);
|
|
int32_t secondaryLen=0;
|
|
unsigned char secondaryBuffer[128];
|
|
if(secondaryEncoderEnabled && secondaryEncoder){
|
|
secondaryLen=opus_encode(secondaryEncoder, data, static_cast<int>(len), secondaryBuffer, sizeof(secondaryBuffer));
|
|
//LOGV("secondaryLen %d", secondaryLen);
|
|
}
|
|
InvokeCallback(buffer, (size_t)r, secondaryBuffer, (size_t)secondaryLen);
|
|
}
|
|
}
|
|
|
|
size_t tgvoip::OpusEncoder::Callback(unsigned char *data, size_t len, void* param){
|
|
OpusEncoder* e=(OpusEncoder*)param;
|
|
unsigned char* buf=e->bufferPool.Get();
|
|
if(buf){
|
|
assert(len==960*2);
|
|
memcpy(buf, data, 960*2);
|
|
e->queue.Put(buf);
|
|
}else{
|
|
LOGW("opus_encoder: no buffer slots left");
|
|
if(e->complexity>1){
|
|
e->complexity--;
|
|
opus_encoder_ctl(e->enc, OPUS_SET_COMPLEXITY(e->complexity));
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
|
|
uint32_t tgvoip::OpusEncoder::GetBitrate(){
|
|
return requestedBitrate;
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::SetEchoCanceller(EchoCanceller* aec){
|
|
echoCanceller=aec;
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::RunThread(){
|
|
uint32_t bufferedCount=0;
|
|
uint32_t packetsPerFrame=frameDuration/20;
|
|
LOGV("starting encoder, packets per frame=%d", packetsPerFrame);
|
|
int16_t* frame;
|
|
if(packetsPerFrame>1)
|
|
frame=(int16_t*) malloc(960*2*packetsPerFrame);
|
|
else
|
|
frame=NULL;
|
|
bool frameHasVoice=false;
|
|
bool wasVadMode=false;
|
|
while(running){
|
|
int16_t* packet=(int16_t*)queue.GetBlocking();
|
|
if(packet){
|
|
bool hasVoice=true;
|
|
if(echoCanceller)
|
|
echoCanceller->ProcessInput(packet, 960, hasVoice);
|
|
if(!postProcEffects.empty()){
|
|
for(effects::AudioEffect* effect:postProcEffects){
|
|
effect->Process(packet, 960);
|
|
}
|
|
}
|
|
if(packetsPerFrame==1){
|
|
Encode(packet, 960);
|
|
}else{
|
|
memcpy(frame+(960*bufferedCount), packet, 960*2);
|
|
frameHasVoice=frameHasVoice || hasVoice;
|
|
bufferedCount++;
|
|
if(bufferedCount==packetsPerFrame){
|
|
if(vadMode){
|
|
if(frameHasVoice){
|
|
opus_encoder_ctl(enc, OPUS_SET_BITRATE(currentBitrate));
|
|
opus_encoder_ctl(enc, OPUS_SET_BANDWIDTH(vadModeVoiceBandwidth));
|
|
if(secondaryEncoder){
|
|
opus_encoder_ctl(secondaryEncoder, OPUS_SET_BITRATE(currentBitrate));
|
|
opus_encoder_ctl(secondaryEncoder, OPUS_SET_BANDWIDTH(vadModeVoiceBandwidth));
|
|
}
|
|
}else{
|
|
opus_encoder_ctl(enc, OPUS_SET_BITRATE(vadNoVoiceBitrate));
|
|
opus_encoder_ctl(enc, OPUS_SET_BANDWIDTH(vadModeNoVoiceBandwidth));
|
|
if(secondaryEncoder){
|
|
opus_encoder_ctl(secondaryEncoder, OPUS_SET_BITRATE(vadNoVoiceBitrate));
|
|
opus_encoder_ctl(secondaryEncoder, OPUS_SET_BANDWIDTH(vadModeNoVoiceBandwidth));
|
|
}
|
|
}
|
|
wasVadMode=true;
|
|
}else if(wasVadMode){
|
|
wasVadMode=false;
|
|
opus_encoder_ctl(enc, OPUS_SET_BITRATE(currentBitrate));
|
|
opus_encoder_ctl(enc, OPUS_SET_BANDWIDTH(secondaryEncoderEnabled ? secondaryEnabledBandwidth : OPUS_AUTO));
|
|
if(secondaryEncoder){
|
|
opus_encoder_ctl(secondaryEncoder, OPUS_SET_BITRATE(currentBitrate));
|
|
opus_encoder_ctl(secondaryEncoder, OPUS_SET_BANDWIDTH(secondaryEnabledBandwidth));
|
|
}
|
|
}
|
|
Encode(frame, 960*packetsPerFrame);
|
|
bufferedCount=0;
|
|
frameHasVoice=false;
|
|
}
|
|
}
|
|
bufferPool.Reuse(reinterpret_cast<unsigned char *>(packet));
|
|
}
|
|
}
|
|
if(frame)
|
|
free(frame);
|
|
}
|
|
|
|
|
|
void tgvoip::OpusEncoder::SetOutputFrameDuration(uint32_t duration){
|
|
frameDuration=duration;
|
|
}
|
|
|
|
|
|
void tgvoip::OpusEncoder::SetPacketLoss(int percent){
|
|
packetLossPercent=std::min(20, percent);
|
|
opus_encoder_ctl(enc, OPUS_SET_PACKET_LOSS_PERC(packetLossPercent));
|
|
opus_encoder_ctl(enc, OPUS_SET_INBAND_FEC(percent>0 && !secondaryEncoderEnabled ? 1 : 0));
|
|
}
|
|
|
|
int tgvoip::OpusEncoder::GetPacketLoss(){
|
|
return packetLossPercent;
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::SetDTX(bool enable){
|
|
opus_encoder_ctl(enc, OPUS_SET_DTX(enable ? 1 : 0));
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::SetLevelMeter(tgvoip::AudioLevelMeter *levelMeter){
|
|
this->levelMeter=levelMeter;
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::SetCallback(void (*f)(unsigned char *, size_t, unsigned char *, size_t, void *), void *param){
|
|
callback=f;
|
|
callbackParam=param;
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::InvokeCallback(unsigned char *data, size_t length, unsigned char *secondaryData, size_t secondaryLength){
|
|
callback(data, length, secondaryData, secondaryLength, callbackParam);
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::SetSecondaryEncoderEnabled(bool enabled){
|
|
secondaryEncoderEnabled=enabled;
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::SetVadMode(bool vad){
|
|
vadMode=vad;
|
|
}
|
|
void tgvoip::OpusEncoder::AddAudioEffect(effects::AudioEffect *effect){
|
|
postProcEffects.push_back(effect);
|
|
}
|
|
|
|
void tgvoip::OpusEncoder::RemoveAudioEffect(effects::AudioEffect *effect){
|
|
std::vector<effects::AudioEffect*>::iterator i=std::find(postProcEffects.begin(), postProcEffects.end(), effect);
|
|
if(i!=postProcEffects.end())
|
|
postProcEffects.erase(i);
|
|
}
|