This repository has been archived by the owner on Jan 16, 2024. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 3
/
cod_ld8a.c
471 lines (382 loc) · 20.5 KB
/
cod_ld8a.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
/*
ITU-T G.729A Speech Coder with Annex B ANSI-C Source Code
*/
/*
----------------------------------------------------------------------
COPYRIGHT NOTICE
----------------------------------------------------------------------
ITU-T G.729 Annex C ANSI C source code
Copyright (C) 1998, AT&T, France Telecom, NTT, University of
Sherbrooke. All rights reserved.
----------------------------------------------------------------------
*/
/*-----------------------------------------------------------------*
* Functions coder_ld8a and init_coder_ld8a *
* ~~~~~~~~~~ ~~~~~~~~~~~~~~~ *
* *
* init_coder_ld8a(void); *
* *
* ->Initialization of variables for the coder section. *
* *
* *
* coder_ld8a(Short ana[]); *
* *
* ->Main coder function. *
* *
* *
* Input: *
* *
* 80 speech data should have beee copy to vector new_speech[]. *
* This vector is global and is declared in this function. *
* *
* Ouputs: *
* *
* ana[] ->analysis parameters. *
* *
*-----------------------------------------------------------------*/
#include <math.h>
#include "typedef.h"
#include "ld8a.h"
#include "vad.h"
#include "dtx.h"
#include "sid.h"
#include "tab_ld8a.h"
/*-----------------------------------------------------------*
* Coder constant parameters (defined in "ld8a.h") *
*-----------------------------------------------------------*
* L_WINDOW : LPC analysis window size. *
* L_NEXT : Samples of next frame needed for autocor. *
* L_FRAME : Frame size. *
* L_SUBFR : Sub-frame size. *
* M : LPC order. *
* MP1 : LPC order+1 *
* L_TOTAL : Total size of speech buffer. *
* PIT_MIN : Minimum pitch lag. *
* PIT_MAX : Maximum pitch lag. *
* L_INTERPOL : Length of filter for interpolation *
*-----------------------------------------------------------*/
/*-----------------------------------------------------------------*
* Function init_coder_ld8a *
* ~~~~~~~~~~~~~~~ *
* *
* init_coder_ld8a(void); *
* *
* ->Initialization of variables for the coder section. *
* - initialize pointers to speech buffer *
* - initialize static pointers *
* - set static vectors to zero *
*-----------------------------------------------------------------*/
void init_coder_ld8a(encoder_state *state)
{
/*----------------------------------------------------------------------*
* Initialize pointers to speech vector. *
* *
* *
* |--------------------|-------------|-------------|------------| *
* previous speech sf1 sf2 L_NEXT *
* *
* <---------------- Total speech vector (L_TOTAL) -----------> *
* <---------------- LPC analysis window (L_WINDOW) -----------> *
* | <-- present frame (L_FRAME) --> *
* old_speech | <-- new speech (L_FRAME) --> *
* p_window | | *
* speech | *
* new_speech *
*-----------------------------------------------------------------------*/
state->new_speech = state->old_speech + L_TOTAL - L_FRAME; /* New speech */
state->speech = state->new_speech - L_NEXT; /* Present frame */
state->p_window = state->old_speech + L_TOTAL - L_WINDOW; /* For LPC window */
/* Initialize static pointers */
state->wsp = state->old_wsp + PIT_MAX;
state->exc = state->old_exc + PIT_MAX + L_INTERPOL;
/* Static vectors to zero */
set_zero(state->old_speech, L_TOTAL);
set_zero(state->old_exc, PIT_MAX+L_INTERPOL);
set_zero(state->old_wsp, PIT_MAX);
set_zero(state->mem_w, M);
set_zero(state->mem_w0, M);
set_zero(state->mem_zero, M);
state->sharp = SHARPMIN;
copy(lsp_old, state->lsp_old, M);
copy(state->lsp_old, state->lsp_old_q, M);
lsp_encw_reset(&state->lsp_state);
init_exc_err(state->cng_state.exc_err);
set_zero(state->old_A, M+1);
state->old_A[0]= (F)1.;
set_zero(state->old_rc, 2);
/* For G.729B */
/* Initialize VAD/DTX parameters */
state->seed = INIT_SEED;
state->pastVad = 1;
state->ppastVad = 1;
vad_init(&state->vad_state);
init_lsfq_noise(state->cng_state.noise_fg);
copy(past_qua_en, state->past_qua_en, 4);
return;
}
/*-----------------------------------------------------------------*
* Function coder_ld8a *
* ~~~~~~~~~~ *
* coder_ld8a(int ana[]); *
* *
* ->Main coder function. *
* *
* *
* Input: *
* *
* 80 speech data should have beee copy to vector new_speech[]. *
* This vector is global and is declared in this function. *
* *
* Ouputs: *
* *
* ana[] ->analysis parameters. *
* *
*-----------------------------------------------------------------*/
void coder_ld8a(encoder_state *state,
int ana[], /* output: analysis parameters */
int frame, /* input : frame counter */
int dtx_enable /* input : DTX enable flag */
)
{
/* LPC coefficients */
FLOAT Aq_t[(MP1)*2]; /* A(z) quantized for the 2 subframes */
FLOAT Ap_t[(MP1)*2]; /* A(z) with spectral expansion */
FLOAT *Aq, *Ap; /* Pointer on Aq_t and Ap_t */
/* Other vectors */
FLOAT h1[L_SUBFR]; /* Impulse response h1[] */
FLOAT xn[L_SUBFR]; /* Target vector for pitch search */
FLOAT xn2[L_SUBFR]; /* Target vector for codebook search */
FLOAT code[L_SUBFR]; /* Fixed codebook excitation */
FLOAT y1[L_SUBFR]; /* Filtered adaptive excitation */
FLOAT y2[L_SUBFR]; /* Filtered fixed codebook excitation */
FLOAT g_coeff[5]; /* Correlations between xn, y1, & y2:
<y1,y1>, <xn,y1>, <y2,y2>, <xn,y2>,<y1,y2>*/
/* Scalars */
int i, j, i_subfr;
int T_op, T0, T0_min, T0_max, T0_frac;
int index;
FLOAT gain_pit, gain_code;
int taming;
/*------------------------------------------------------------------------*
* - Perform LPC analysis: *
* * autocorrelation + lag windowing *
* * Levinson-durbin algorithm to find a[] *
* * convert a[] to lsp[] *
* * quantize and code the LSPs *
* * find the interpolated LSPs and convert to a[] for the 2 *
* subframes (both quantized and unquantized) *
*------------------------------------------------------------------------*/
{
/* Temporary vectors */
FLOAT r[NP+1]; /* Autocorrelations */
FLOAT rc[M]; /* Reflexion coefficients */
FLOAT lsp_new[M]; /* lsp coefficients */
FLOAT lsp_new_q[M]; /* Quantized lsp coeff. */
FLOAT lsf_new[M];
/* For G.729B */
FLOAT r_nbe[MP1];
FLOAT lsfq_mem[MA_NP][M];
int Vad;
FLOAT Energy_db;
/* LP analysis */
autocorr(state->p_window, NP, r); /* Autocorrelations */
copy(r, r_nbe, MP1);
lag_window(NP, r); /* Lag windowing */
levinson(r, Ap_t, rc, state->old_A, state->old_rc); /* Levinson Durbin */
az_lsp(Ap_t, lsp_new, state->lsp_old); /* Convert A(z) to lsp */
if (dtx_enable == 1)
{
lsp_lsf(lsp_new, lsf_new, M);
vad(&state->vad_state, rc[1], lsf_new, r, state->p_window, frame,
state->pastVad, state->ppastVad, &Vad, &Energy_db);
update_cng(&state->cng_state, r_nbe, Vad);
} else Vad = 1;
/* ---------------------- */
/* Case of Inactive frame */
/* ---------------------- */
if (Vad == 0)
{
//get_freq_prev(&state->lsp_state, &lsfq_mem[i][0], MA_NP);
for (i=0; i<MA_NP; i++) copy(&state->lsp_state.freq_prev[i][0], &lsfq_mem[i][0], M);
cod_cng(&state->cng_state, state->exc, state->pastVad, state->lsp_old_q, state->old_A, state->old_rc, Aq_t,
ana, lsfq_mem, &state->seed);
//update_freq_prev(&state->lsp_state, &lsfq_mem[i][0], MA_NP);
for (i=0; i<MA_NP; i++) copy(&lsfq_mem[i][0], &state->lsp_state.freq_prev[i][0], M);
state->ppastVad = state->pastVad;
state->pastVad = Vad;
/* Update wsp, mem_w and mem_w0 */
Aq = Aq_t;
for (i_subfr = 0; i_subfr < L_FRAME; i_subfr += L_SUBFR) {
residu(Aq, &state->speech[i_subfr], xn, L_SUBFR);
weight_az(Aq, GAMMA1, M, Ap_t);
/* Compute wsp and mem_w */
Ap = Ap_t + MP1;
Ap[0] = (F)0.125;
for (i = 1; i <= M; i++)
Ap[i] = (F)(Ap_t[i] - (F)0.7 * Ap_t[i-1]);
syn_filt(Ap, xn, &state->wsp[i_subfr], L_SUBFR, state->mem_w, 1);
/* Compute mem_w0 */
for (i = 0; i < L_SUBFR; i++)
xn[i] = xn[i] - state->exc[i_subfr+i]; /* residu[] - exc[] */
syn_filt(Ap_t, xn, xn, L_SUBFR, state->mem_w0, 1);
Aq += MP1;
}
state->sharp = SHARPMIN;
/* Update memories for next frames */
copy(&state->old_speech[L_FRAME], &state->old_speech[0], L_TOTAL-L_FRAME);
copy(&state->old_wsp[L_FRAME], &state->old_wsp[0], PIT_MAX);
copy(&state->old_exc[L_FRAME], &state->old_exc[0], PIT_MAX+L_INTERPOL);
return;
} /* End of inactive frame case */
/* -------------------- */
/* Case of Active frame */
/* -------------------- */
/* Case of active frame */
*ana++ = 1;
state->seed = INIT_SEED;
state->ppastVad = state->pastVad;
state->pastVad = Vad;
/* LSP quantization */
qua_lsp(&state->lsp_state, lsp_new, lsp_new_q, ana);
ana += 2; /* Advance analysis parameters pointer */
/*--------------------------------------------------------------------*
* Find interpolated LPC parameters in all subframes *
* The interpolated parameters are in array Aq_t[]. *
*--------------------------------------------------------------------*/
int_qlpc(state->lsp_old_q, lsp_new_q, Aq_t);
/* Compute A(z/gamma) */
weight_az(&Aq_t[0], GAMMA1, M, &Ap_t[0]);
weight_az(&Aq_t[MP1], GAMMA1, M, &Ap_t[MP1]);
/* update the LSPs for the next frame */
copy(lsp_new, state->lsp_old, M);
copy(lsp_new_q, state->lsp_old_q, M);
}
/*----------------------------------------------------------------------*
* - Find the weighted input speech w_sp[] for the whole speech frame *
* - Find the open-loop pitch delay for the whole speech frame *
* - Set the range for searching closed-loop pitch in 1st subframe *
*----------------------------------------------------------------------*/
residu(&Aq_t[0], &state->speech[0], &state->exc[0], L_SUBFR);
residu(&Aq_t[MP1], &state->speech[L_SUBFR], &state->exc[L_SUBFR], L_SUBFR);
{
FLOAT Ap1[MP1];
Ap = Ap_t;
Ap1[0] = (F)1.0;
for(i=1; i<=M; i++)
Ap1[i] = Ap[i] - (F)0.7 * Ap[i-1];
syn_filt(Ap1, &state->exc[0], &state->wsp[0], L_SUBFR, state->mem_w, 1);
Ap += MP1;
for(i=1; i<=M; i++)
Ap1[i] = Ap[i] - (F)0.7 * Ap[i-1];
syn_filt(Ap1, &state->exc[L_SUBFR], &state->wsp[L_SUBFR], L_SUBFR, state->mem_w, 1);
}
/* Find open loop pitch lag for whole speech frame */
T_op = pitch_ol_fast(state->wsp, L_FRAME);
/* Range for closed loop pitch search in 1st subframe */
T0_min = T_op - 3;
if (T0_min < PIT_MIN) T0_min = PIT_MIN;
T0_max = T0_min + 6;
if (T0_max > PIT_MAX)
{
T0_max = PIT_MAX;
T0_min = T0_max - 6;
}
/*------------------------------------------------------------------------*
* Loop for every subframe in the analysis frame *
*------------------------------------------------------------------------*
* To find the pitch and innovation parameters. The subframe size is *
* L_SUBFR and the loop is repeated L_FRAME/L_SUBFR times. *
* - find the weighted LPC coefficients *
* - find the LPC residual signal *
* - compute the target signal for pitch search *
* - compute impulse response of weighted synthesis filter (h1[]) *
* - find the closed-loop pitch parameters *
* - encode the pitch delay *
* - find target vector for codebook search *
* - codebook search *
* - VQ of pitch and codebook gains *
* - update states of weighting filter *
*------------------------------------------------------------------------*/
Aq = Aq_t; /* pointer to interpolated quantized LPC parameters */
Ap = Ap_t; /* pointer to weighted LPC coefficients */
for (i_subfr = 0; i_subfr < L_FRAME; i_subfr += L_SUBFR)
{
/*---------------------------------------------------------------*
* Compute impulse response, h1[], of weighted synthesis filter *
*---------------------------------------------------------------*/
h1[0] = (F)1.0;
set_zero(&h1[1], L_SUBFR-1);
syn_filt(Ap, h1, h1, L_SUBFR, &h1[1], 0);
/*-----------------------------------------------*
* Find the target vector for pitch search: *
*----------------------------------------------*/
syn_filt(Ap, &state->exc[i_subfr], xn, L_SUBFR, state->mem_w0, 0);
/*-----------------------------------------------------------------*
* Closed-loop fractional pitch search *
*-----------------------------------------------------------------*/
T0 = pitch_fr3_fast(&state->exc[i_subfr], xn, h1, L_SUBFR, T0_min, T0_max,
i_subfr, &T0_frac);
index = enc_lag3(T0, T0_frac, &T0_min, &T0_max, PIT_MIN, PIT_MAX,
i_subfr);
*ana++ = index;
if (i_subfr == 0)
*ana++ = parity_pitch(index);
/*-----------------------------------------------------------------*
* - find filtered pitch exc *
* - compute pitch gain and limit between 0 and 1.2 *
* - update target vector for codebook search *
* - find LTP residual. *
*-----------------------------------------------------------------*/
syn_filt(Ap, &state->exc[i_subfr], y1, L_SUBFR, state->mem_zero, 0);
gain_pit = g_pitch(xn, y1, g_coeff, L_SUBFR);
/* clip pitch gain if taming is necessary */
taming = test_err(state->cng_state.exc_err, T0, T0_frac);
if( taming == 1){
if (gain_pit > GPCLIP) {
gain_pit = GPCLIP;
}
}
for (i = 0; i < L_SUBFR; i++)
xn2[i] = xn[i] - y1[i]*gain_pit;
/*-----------------------------------------------------*
* - Innovative codebook search. *
*-----------------------------------------------------*/
index = ACELP_code_A(xn2, h1, T0, state->sharp, code, y2, &i);
*ana++ = index; /* Positions index */
*ana++ = i; /* Signs index */
/*------------------------------------------------------*
* - Compute the correlations <y2,y2>, <xn,y2>, <y1,y2>*
* - Vector quantize gains. *
*------------------------------------------------------*/
corr_xy2(xn, y1, y2, g_coeff);
*ana++ =qua_gain(state->past_qua_en, code, g_coeff, L_SUBFR, &gain_pit, &gain_code,
taming);
/*------------------------------------------------------------*
* - Update pitch sharpening "sharp" with quantized gain_pit *
*------------------------------------------------------------*/
state->sharp = gain_pit;
if (state->sharp > SHARPMAX) state->sharp = SHARPMAX;
if (state->sharp < SHARPMIN) state->sharp = SHARPMIN;
/*------------------------------------------------------*
* - Find the total excitation *
* - update filters' memories for finding the target *
* vector in the next subframe (mem_w0[]) *
*------------------------------------------------------*/
for (i = 0; i < L_SUBFR; i++)
state->exc[i+i_subfr] = gain_pit*state->exc[i+i_subfr] + gain_code*code[i];
update_exc_err(state->cng_state.exc_err, gain_pit, T0);
for (i = L_SUBFR-M, j = 0; i < L_SUBFR; i++, j++)
state->mem_w0[j] = xn[i] - gain_pit*y1[i] - gain_code*y2[i];
Aq += MP1; /* interpolated LPC parameters for next subframe */
Ap += MP1;
}
/*--------------------------------------------------*
* Update signal for next frame. *
* -> shift to the left by L_FRAME: *
* speech[], wsp[] and exc[] *
*--------------------------------------------------*/
copy(&state->old_speech[L_FRAME], &state->old_speech[0], L_TOTAL-L_FRAME);
copy(&state->old_wsp[L_FRAME], &state->old_wsp[0], PIT_MAX);
copy(&state->old_exc[L_FRAME], &state->old_exc[0], PIT_MAX+L_INTERPOL);
return;
}