static int tdav_speex_jitterbuffer_open(tmedia_jitterbuffer_t* self, uint32_t frame_duration, uint32_t rate, uint32_t channels)
{
tdav_speex_jitterbuffer_t *jitterbuffer = (tdav_speex_jitterbuffer_t *)self;
spx_int32_t tmp;
TSK_DEBUG_INFO("Open speex jb (ptime=%u, rate=%u)", frame_duration, rate);
if(!(jitterbuffer->state = jitter_buffer_init((int)frame_duration))){
TSK_DEBUG_ERROR("jitter_buffer_init() failed");
return -2;
}
jitterbuffer->rate = rate;
jitterbuffer->frame_duration = frame_duration;
jitterbuffer->channels = channels;
jitterbuffer->x_data_size = ((frame_duration * jitterbuffer->rate) / 500) << (channels == 2 ? 1 : 0);
jitter_buffer_ctl(jitterbuffer->state, JITTER_BUFFER_GET_MARGIN, &tmp);
TSK_DEBUG_INFO("Default Jitter buffer margin=%d", tmp);
jitter_buffer_ctl(jitterbuffer->state, JITTER_BUFFER_GET_MAX_LATE_RATE, &tmp);
TSK_DEBUG_INFO("Default Jitter max late rate=%d", tmp);
if((tmp = tmedia_defaults_get_jb_margin()) >= 0){
jitter_buffer_ctl(jitterbuffer->state, JITTER_BUFFER_SET_MARGIN, &tmp);
TSK_DEBUG_INFO("New Jitter buffer margin=%d", tmp);
}
if((tmp = tmedia_defaults_get_jb_max_late_rate()) >= 0){
jitter_buffer_ctl(jitterbuffer->state, JITTER_BUFFER_SET_MAX_LATE_RATE, &tmp);
TSK_DEBUG_INFO("New Jitter buffer max late rate=%d", tmp);
}
return 0;
}
int maintestjitter()
{
char buffer[65536];
JitterBufferPacket in, out;
int i;
JitterBuffer *jb = jitter_buffer_init(10);
out.data = buffer;
/* Frozen sender case */
jitterFill(jb);
for(i=0;i<100;++i) {
out.len = 65536;
jitter_buffer_get(jb, &out, 10, NULL);
jitter_buffer_tick(jb);
}
synthIn(&in, 100, 1);
jitter_buffer_put(jb, &in);
out.len = 65536;
if (jitter_buffer_get(jb, &out, 10, NULL) != JITTER_BUFFER_OK) {
printf("Failed frozen sender resynchronize\n");
} else {
printf("Frozen sender: Jitter %d\n", out.timestamp - 100*10);
}
return 0;
}
JNIEXPORT jlong JNICALL Native_NATIVE(jitter_1buffer_1init)
(JNIEnv *env, jclass that, jint arg0)
{
jlong rc = 0;
Native_NATIVE_ENTER(env, that, Native_jitter_1buffer_1init_FUNC);
rc = (intptr_t)(JitterBuffer *)jitter_buffer_init(arg0);
Native_NATIVE_EXIT(env, that, Native_jitter_1buffer_1init_FUNC);
return rc;
}
void speex_jitter_init(SpeexJitter *jitter, void *decoder, int sampling_rate)
{
jitter->dec = decoder;
speex_decoder_ctl(decoder, SPEEX_GET_FRAME_SIZE, &jitter->frame_size);
jitter->packets = jitter_buffer_init(jitter->frame_size);
speex_bits_init(&jitter->current_packet);
jitter->valid_bits = 0;
jitter->activity_threshold = 0;
}
AudioOutputSpeech::AudioOutputSpeech(ClientUser *user, unsigned int freq, MessageHandler::UDPMessageType type) : AudioOutputUser(user->qsName) {
int err;
p = user;
umtType = type;
iMixerFreq = freq;
cCodec = NULL;
cdDecoder = NULL;
dsSpeex = NULL;
opusState = NULL;
bStereo = false;
iSampleRate = SAMPLE_RATE;
iFrameSize = iSampleRate / 100;
iAudioBufferSize = iFrameSize;
if (umtType == MessageHandler::UDPVoiceOpus) {
#ifdef USE_OPUS
iAudioBufferSize *= 12;
opusState = opus_decoder_create(iSampleRate, bStereo ? 2 : 1, NULL);
#endif
} else if (umtType == MessageHandler::UDPVoiceSpeex) {
speex_bits_init(&sbBits);
dsSpeex = speex_decoder_init(speex_lib_get_mode(SPEEX_MODEID_UWB));
int iArg=1;
speex_decoder_ctl(dsSpeex, SPEEX_SET_ENH, &iArg);
speex_decoder_ctl(dsSpeex, SPEEX_GET_FRAME_SIZE, &iFrameSize);
speex_decoder_ctl(dsSpeex, SPEEX_GET_SAMPLING_RATE, &iSampleRate);
iAudioBufferSize = iFrameSize;
}
iOutputSize = static_cast<unsigned int>(ceilf(static_cast<float>(iAudioBufferSize * iMixerFreq) / static_cast<float>(iSampleRate)));
if (bStereo) {
iAudioBufferSize *= 2;
iOutputSize *= 2;
}
srs = NULL;
fResamplerBuffer = NULL;
if (iMixerFreq != iSampleRate) {
srs = speex_resampler_init(bStereo ? 2 : 1, iSampleRate, iMixerFreq, 3, &err);
fResamplerBuffer = new float[iAudioBufferSize];
}
iBufferOffset = iBufferFilled = iLastConsume = 0;
bLastAlive = true;
iMissCount = 0;
iMissedFrames = 0;
ucFlags = 0xFF;
jbJitter = jitter_buffer_init(iFrameSize);
int margin = g.s.iJitterBufferSize * iFrameSize;
jitter_buffer_ctl(jbJitter, JITTER_BUFFER_SET_MARGIN, &margin);
fFadeIn = new float[iFrameSize];
fFadeOut = new float[iFrameSize];
float mul = static_cast<float>(M_PI / (2.0 * static_cast<double>(iFrameSize)));
for (unsigned int i=0;i<iFrameSize;++i)
fFadeIn[i] = fFadeOut[iFrameSize-i-1] = sinf(static_cast<float>(i) * mul);
}
int main(int argc, char **argv)
{
int sd, rc, n, tmp;
char msg[MAX_MSG];
int nfds;
int send_timestamp = 0;
int recv_started = 0;
struct pollfd *pfds;
struct alsa_dev *dev;
CELTEncoder *enc_state;
CELTDecoder *dec_state;
CELTMode *mode;
struct sched_param param;
JitterBuffer *jitter;
SpeexEchoState *echo_state;
char mac_own[6], mac_remote[6];
if (argc != 4)
panic("Usage %s plughw:0,0 <lmac in xx:xx:xx:xx:xx:xx> <rmac>\n", argv[0]);
register_signal(SIGINT, sighandler);
hack_mac(mac_own, argv[2], strlen(argv[2]));
hack_mac(mac_remote, argv[3], strlen(argv[3]));
sd = socket(AF_LANA, SOCK_RAW, 0);
if (sd < 0)
panic("%s: cannot open socket \n", argv[0]);
printf("If ready hit key!\n"); //user must do binding
getchar();
dev = alsa_open(argv[1], SAMPLING_RATE, CHANNELS, FRAME_SIZE);
mode = celt_mode_create(SAMPLING_RATE, FRAME_SIZE, NULL);
enc_state = celt_encoder_create(mode, CHANNELS, NULL);
dec_state = celt_decoder_create(mode, CHANNELS, NULL);
param.sched_priority = sched_get_priority_min(SCHED_FIFO);
if (sched_setscheduler(0, SCHED_FIFO, ¶m))
whine("sched_setscheduler error!\n");
/* Setup all file descriptors for poll()ing */
nfds = alsa_nfds(dev);
pfds = xmalloc(sizeof(*pfds) * (nfds + 1));
alsa_getfds(dev, pfds, nfds);
pfds[nfds].fd = sd;
pfds[nfds].events = POLLIN;
/* Setup jitter buffer using decoder */
jitter = jitter_buffer_init(FRAME_SIZE);
tmp = FRAME_SIZE;
jitter_buffer_ctl(jitter, JITTER_BUFFER_SET_MARGIN, &tmp);
/* Echo canceller with 200 ms tail length */
echo_state = speex_echo_state_init(FRAME_SIZE, 10 * FRAME_SIZE);
tmp = SAMPLING_RATE;
speex_echo_ctl(echo_state, SPEEX_ECHO_SET_SAMPLING_RATE, &tmp);
register_signal_f(SIGALRM, timer_elapsed, SA_SIGINFO);
alsa_start(dev);
printf("ALSA started!\n");
itimer.it_interval.tv_sec = 0;
itimer.it_interval.tv_usec = interval;
itimer.it_value.tv_sec = 0;
itimer.it_value.tv_usec = interval;
setitimer(ITIMER_REAL, &itimer, NULL);
while (!sigint) {
poll(pfds, nfds + 1, -1);
/* Received packets */
if (pfds[nfds].revents & POLLIN) {
memset(msg, 0, MAX_MSG);
n = recv(sd, msg, MAX_MSG, 0);
if (n <= 0)
goto do_alsa;
pkts_in++;
int recv_timestamp;
memcpy(&recv_timestamp, msg, sizeof(recv_timestamp));
JitterBufferPacket packet;
packet.data = msg+4/*+6+6+2*/;
packet.len = n-4/*-6-6-2*/;
packet.timestamp = recv_timestamp;
packet.span = FRAME_SIZE;
packet.sequence = 0;
/* Put content of the packet into the jitter buffer,
except for the pseudo-header */
jitter_buffer_put(jitter, &packet);
recv_started = 1;
}
do_alsa:
/* Ready to play a frame (playback) */
if (alsa_play_ready(dev, pfds, nfds)) {
short pcm[FRAME_SIZE * CHANNELS] = {0};
if (recv_started) {
JitterBufferPacket packet;
/* Get audio from the jitter buffer */
packet.data = msg;
packet.len = MAX_MSG;
jitter_buffer_tick(jitter);
jitter_buffer_get(jitter, &packet, FRAME_SIZE,
NULL);
if (packet.len == 0)
packet.data=NULL;
celt_decode(dec_state, (const unsigned char *)
packet.data, packet.len, pcm);
}
/* Playback the audio and reset the echo canceller
if we got an underrun */
alsa_write(dev, pcm, FRAME_SIZE);
// if (alsa_write(dev, pcm, FRAME_SIZE))
// speex_echo_state_reset(echo_state);
/* Put frame into playback buffer */
// speex_echo_playback(echo_state, pcm);
}
/* Audio available from the soundcard (capture) */
if (alsa_cap_ready(dev, pfds, nfds)) {
short pcm[FRAME_SIZE * CHANNELS];
//pcm2[FRAME_SIZE * CHANNELS];
char outpacket[MAX_MSG];
alsa_read(dev, pcm, FRAME_SIZE);
/* Perform echo cancellation */
// speex_echo_capture(echo_state, pcm, pcm2);
// for (i = 0; i < FRAME_SIZE * CHANNELS; ++i)
// pcm[i] = pcm2[i];
celt_encode(enc_state, pcm, NULL, (unsigned char *)
(outpacket+4+6+6+2), PACKETSIZE);
/* Pseudo header: four null bytes and a 32-bit
timestamp; XXX hack */
memcpy(outpacket,mac_remote,6);
memcpy(outpacket+6,mac_own,6);
outpacket[6+6] = (uint8_t) 0xac;
outpacket[6+6+1] = (uint8_t) 0xdc;
memcpy(outpacket+6+6+2, &send_timestamp, sizeof(send_timestamp));
send_timestamp += FRAME_SIZE;
rc = sendto(sd, outpacket, PACKETSIZE+4+6+6+2, 0, NULL, 0);
if (rc < 0)
panic("cannot send to socket");
pkts_out++;
}
}
itimer.it_interval.tv_sec = 0;
itimer.it_interval.tv_usec = 0;
itimer.it_value.tv_sec = 0;
itimer.it_value.tv_usec = 0;
setitimer(ITIMER_REAL, &itimer, NULL);
close(sd);
return 0;
}
AudioInput::AudioInput() {
speex_bits_init(&sbBits);
speex_bits_reset(&sbBits);
iFrames = 0;
iSampleRate = SAMPLE_RATE;
esEncState=speex_encoder_init(speex_lib_get_mode(SPEEX_MODEID_WB));
speex_encoder_ctl(esEncState,SPEEX_GET_FRAME_SIZE,&iFrameSize);
mumble_drft_init(&fftTable, iFrameSize);
iFrameCounter = 0;
iSilentFrames = 0;
iHoldFrames = 0;
iBestBin = 0;
int iArg=1;
float fArg=0.0;
speex_encoder_ctl(esEncState,SPEEX_SET_VBR, &iArg);
iArg = 0;
speex_encoder_ctl(esEncState,SPEEX_SET_VAD, &iArg);
speex_encoder_ctl(esEncState,SPEEX_SET_DTX, &iArg);
fArg = static_cast<float>(g.s.iQuality);
speex_encoder_ctl(esEncState,SPEEX_SET_VBR_QUALITY, &fArg);
speex_encoder_ctl(esEncState,SPEEX_GET_BITRATE, &iArg);
speex_encoder_ctl(esEncState, SPEEX_SET_VBR_MAX_BITRATE, &iArg);
iArg = 5;
speex_encoder_ctl(esEncState,SPEEX_SET_COMPLEXITY, &iArg);
bResetProcessor = true;
sppPreprocess = NULL;
sesEcho = NULL;
srsMic = srsEcho = NULL;
jb = jitter_buffer_init(10);
iJitterSeq = 1;
psMic = new short[iFrameSize];
psSpeaker = new short[iFrameSize];
psClean = new short[iFrameSize];
iEchoChannels = iMicChannels = 0;
iEchoFreq = iMicFreq = SAMPLE_RATE;
iEchoFilled = iMicFilled = 0;
eMicFormat = eEchoFormat = SampleFloat;
iMicSampleSize = iEchoSampleSize = 0;
bPreviousVoice = false;
pfMicInput = pfEchoInput = pfOutput = NULL;
iBitrate = 0;
dPeakMic = dPeakSignal = dPeakSpeaker = 0.0;
if (g.uiSession) {
setMaxBandwidth(g.iMaxBandwidth);
}
bRunning = false;
connect(this, SIGNAL(doMute()), g.mw->qaAudioMute, SLOT(trigger()), Qt::QueuedConnection);
}
AudioOutputSpeech::AudioOutputSpeech(boost::shared_ptr<ClientUser> user, MessageHandler::UDPMessageType type) : AudioOutputUser(user->qsName) {
int err;
p = user;
umtType = type;
unsigned int srate = 0;
cCodec = NULL;
cdDecoder = NULL;
if (umtType != MessageHandler::UDPVoiceSpeex) {
srate = SAMPLE_RATE;
iFrameSize = srate / 100;
dsSpeex = NULL;
} else {
speex_bits_init(&sbBits);
dsSpeex = speex_decoder_init(speex_lib_get_mode(SPEEX_MODEID_UWB));
int iArg=1;
speex_decoder_ctl(dsSpeex, SPEEX_SET_ENH, &iArg);
speex_decoder_ctl(dsSpeex, SPEEX_GET_FRAME_SIZE, &iFrameSize);
speex_decoder_ctl(dsSpeex, SPEEX_GET_SAMPLING_RATE, &srate);
}
if (srate != SAMPLE_RATE)
srs = speex_resampler_init(1, srate, SAMPLE_RATE, 3, &err);
else
srs = NULL;
iOutputSize = static_cast<unsigned int>(ceilf(static_cast<float>(iFrameSize * SAMPLE_RATE) / static_cast<float>(SAMPLE_RATE)));
iBufferOffset = iBufferFilled = iLastConsume = 0;
bLastAlive = true;
iMissCount = 0;
iMissedFrames = 0;
ucFlags = 0xFF;
jbJitter = jitter_buffer_init(iFrameSize);
int margin = g.s.iJitterBufferSize * iFrameSize;
jitter_buffer_ctl(jbJitter, JITTER_BUFFER_SET_MARGIN, &margin);
fFadeIn = new float[iFrameSize];
fFadeOut = new float[iFrameSize];
float mul = static_cast<float>(M_PI / (2.0 * static_cast<double>(iFrameSize)));
for (unsigned int i=0;i<iFrameSize;++i)
fFadeIn[i] = fFadeOut[iFrameSize-i-1] = sinf(static_cast<float>(i) * mul);
}
