eden-emu/opus
14
0
Fork 0
mirror of https://github.com/xiph/opus.git synced 2025-05-16 08:28:29 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

dump_data: remove all noise and output one signal

Browse source 
Now used for fargan instead of LPCNet
This commit is contained in:
Jean-Marc Valin 2024-04-09 14:29:12 -04:00
parent 4c5c498825
commit fdb198e886
No known key found for this signature in database
GPG key ID: 5E5DD9A36F9189C8
1 changed files with 1 additions and 43 deletions
Download patch file Download diff file Expand all files Collapse all files

44
dnn/dump_data.c
View file

@ -68,13 +68,6 @@ static void rand_resp(float *a, float *b) {
b[1] = .75*uni_rand(); b[1] = .75*uni_rand();
} }
void compute_noise(int *noise, float noise_std) {
int i;
for (i=0;i<FRAME_SIZE;i++) {
noise[i] = (int)floor(.5 + noise_std*.707*(log_approx(rand()/(float)RAND_MAX)-log_approx(rand()/(float)RAND_MAX)));
}
}
static opus_int16 float2short(float x) static opus_int16 float2short(float x)
{ {
int i; int i;
@ -82,30 +75,6 @@ static opus_int16 float2short(float x)
return IMAX(-32767, IMIN(32767, i)); return IMAX(-32767, IMIN(32767, i));
} }
void write_audio(LPCNetEncState *st, const opus_int16 *pcm, const int *noise, FILE *file) {
int i;
opus_int16 data[2*FRAME_SIZE];
for (i=0;i<FRAME_SIZE;i++) {
float p=0;
float e;
int j;
for (j=0;j<LPC_ORDER;j++) p -= st->features[NB_BANDS+2+j]*st->sig_mem[j];
e = lin2ulaw(pcm[i] - p);
/* Signal in. */
data[2*i] = float2short(st->sig_mem[0]);
/* Signal out. */
data[2*i+1] = pcm[i];
/* Simulate error on excitation. */
e += noise[i];
e = IMIN(255, IMAX(0, e));
OPUS_MOVE(&st->sig_mem[1], &st->sig_mem[0], LPC_ORDER-1);
st->sig_mem[0] = p + ulaw2lin(e);
}
fwrite(data, 4*FRAME_SIZE, 1, file);
}
int main(int argc, char **argv) { int main(int argc, char **argv) {
int i; int i;
char *argv0; char *argv0;
@ -123,13 +92,11 @@ int main(int argc, char **argv) {
FILE *ffeat; FILE *ffeat;
FILE *fpcm=NULL; FILE *fpcm=NULL;
opus_int16 pcm[FRAME_SIZE]={0}; opus_int16 pcm[FRAME_SIZE]={0};
int noisebuf[FRAME_SIZE]={0};
opus_int16 tmp[FRAME_SIZE] = {0}; opus_int16 tmp[FRAME_SIZE] = {0};
float speech_gain=1; float speech_gain=1;
float old_speech_gain = 1; float old_speech_gain = 1;
int one_pass_completed = 0; int one_pass_completed = 0;
LPCNetEncState *st; LPCNetEncState *st;
float noise_std=0;
int training = -1; int training = -1;
int burg = 0; int burg = 0;
int pitch = 0; int pitch = 0;
@ -202,16 +169,12 @@ int main(int argc, char **argv) {
for (i=0;i<FRAME_SIZE;i++) x[i] = tmp[i]; for (i=0;i<FRAME_SIZE;i++) x[i] = tmp[i];
if (count*FRAME_SIZE_5MS>=10000000 && one_pass_completed) break; if (count*FRAME_SIZE_5MS>=10000000 && one_pass_completed) break;
if (training && ++gain_change_count > 2821) { if (training && ++gain_change_count > 2821) {
float tmp1, tmp2;
speech_gain = pow(10., (-30+(rand()%40))/20.); speech_gain = pow(10., (-30+(rand()%40))/20.);
if (rand()&1) speech_gain = -speech_gain; if (rand()&1) speech_gain = -speech_gain;
if (rand()%20==0) speech_gain *= .01; if (rand()%20==0) speech_gain *= .01;
if (!pitch && rand()%100==0) speech_gain = 0; if (!pitch && rand()%100==0) speech_gain = 0;
gain_change_count = 0; gain_change_count = 0;
rand_resp(a_sig, b_sig); rand_resp(a_sig, b_sig);
tmp1 = rand()/(float)RAND_MAX;
tmp2 = rand()/(float)RAND_MAX;
noise_std = ABS16(-1.5*log(1e-4+tmp1)-.5*log(1e-4+tmp2));
if (fnoise != NULL) { if (fnoise != NULL) {
long pos; long pos;
/* Randomize the fraction because rand() only gives us 31 bits. */ /* Randomize the fraction because rand() only gives us 31 bits. */
@ -244,15 +207,10 @@ int main(int argc, char **argv) {
fwrite(ceps, sizeof(float), 2*NB_BANDS, ffeat); fwrite(ceps, sizeof(float), 2*NB_BANDS, ffeat);
} }
preemphasis(x, &mem_preemph, x, PREEMPHASIS, FRAME_SIZE); preemphasis(x, &mem_preemph, x, PREEMPHASIS, FRAME_SIZE);
for (i=0;i<FRAME_SIZE;i++) x[i] += rand()/(float)RAND_MAX - .5f;
/* PCM is delayed by 1/2 frame to make the features centered on the frames. */ /* PCM is delayed by 1/2 frame to make the features centered on the frames. */
for (i=0;i<FRAME_SIZE-TRAINING_OFFSET;i++) pcm[i+TRAINING_OFFSET] = float2short(x[i]); for (i=0;i<FRAME_SIZE-TRAINING_OFFSET;i++) pcm[i+TRAINING_OFFSET] = float2short(x[i]);
compute_frame_features(st, x, arch); compute_frame_features(st, x, arch);
if (fpcm) {
compute_noise(noisebuf, noise_std);
}
if (pitch) { if (pitch) {
signed char pitch_features[PITCH_MAX_PERIOD-PITCH_MIN_PERIOD+PITCH_IF_FEATURES]; signed char pitch_features[PITCH_MAX_PERIOD-PITCH_MIN_PERIOD+PITCH_IF_FEATURES];
for (i=0;i<PITCH_MAX_PERIOD-PITCH_MIN_PERIOD;i++) { for (i=0;i<PITCH_MAX_PERIOD-PITCH_MIN_PERIOD;i++) {
@ -266,7 +224,7 @@ int main(int argc, char **argv) {
fwrite(st->features, sizeof(float), NB_TOTAL_FEATURES, ffeat); fwrite(st->features, sizeof(float), NB_TOTAL_FEATURES, ffeat);
} }
/*if(pitch) fwrite(pcm, FRAME_SIZE, 2, stdout);*/ /*if(pitch) fwrite(pcm, FRAME_SIZE, 2, stdout);*/
if (fpcm) write_audio(st, pcm, noisebuf, fpcm); if (fpcm) fwrite(pcm, FRAME_SIZE, 2, fpcm);
/*if (fpcm) fwrite(pcm, sizeof(opus_int16), FRAME_SIZE, fpcm);*/ /*if (fpcm) fwrite(pcm, sizeof(opus_int16), FRAME_SIZE, fpcm);*/
for (i=0;i<TRAINING_OFFSET;i++) pcm[i] = float2short(x[i+FRAME_SIZE-TRAINING_OFFSET]); for (i=0;i<TRAINING_OFFSET;i++) pcm[i] = float2short(x[i+FRAME_SIZE-TRAINING_OFFSET]);
old_speech_gain = speech_gain; old_speech_gain = speech_gain;