1 /*****************************************************************************
3 *****************************************************************************
4 * Copyright (C) 1999, 2000 VideoLAN
6 * Authors: Michel Kaempf <maxx@via.ecp.fr>
7 * Aaron Holtzman <aholtzma@engr.uvic.ca>
8 * Renaud Dartus <reno@videolan.org>
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111, USA.
23 *****************************************************************************/
30 #include "int_types.h"
31 #include "ac3_decoder.h"
32 #include "ac3_internal.h"
34 #include "ac3_downmix.h"
36 void imdct_do_256(imdct_t * p_imdct, float x[],float y[], int id);
37 void imdct_do_512(imdct_t * p_imdct, float x[],float y[], int id);
39 /* 128 point bit-reverse LUT */
40 static const u8 bit_reverse_512[] = {
41 0x00, 0x40, 0x20, 0x60, 0x10, 0x50, 0x30, 0x70,
42 0x08, 0x48, 0x28, 0x68, 0x18, 0x58, 0x38, 0x78,
43 0x04, 0x44, 0x24, 0x64, 0x14, 0x54, 0x34, 0x74,
44 0x0c, 0x4c, 0x2c, 0x6c, 0x1c, 0x5c, 0x3c, 0x7c,
45 0x02, 0x42, 0x22, 0x62, 0x12, 0x52, 0x32, 0x72,
46 0x0a, 0x4a, 0x2a, 0x6a, 0x1a, 0x5a, 0x3a, 0x7a,
47 0x06, 0x46, 0x26, 0x66, 0x16, 0x56, 0x36, 0x76,
48 0x0e, 0x4e, 0x2e, 0x6e, 0x1e, 0x5e, 0x3e, 0x7e,
49 0x01, 0x41, 0x21, 0x61, 0x11, 0x51, 0x31, 0x71,
50 0x09, 0x49, 0x29, 0x69, 0x19, 0x59, 0x39, 0x79,
51 0x05, 0x45, 0x25, 0x65, 0x15, 0x55, 0x35, 0x75,
52 0x0d, 0x4d, 0x2d, 0x6d, 0x1d, 0x5d, 0x3d, 0x7d,
53 0x03, 0x43, 0x23, 0x63, 0x13, 0x53, 0x33, 0x73,
54 0x0b, 0x4b, 0x2b, 0x6b, 0x1b, 0x5b, 0x3b, 0x7b,
55 0x07, 0x47, 0x27, 0x67, 0x17, 0x57, 0x37, 0x77,
56 0x0f, 0x4f, 0x2f, 0x6f, 0x1f, 0x5f, 0x3f, 0x7f};
58 static const u8 bit_reverse_256[] = {
59 0x00, 0x20, 0x10, 0x30, 0x08, 0x28, 0x18, 0x38,
60 0x04, 0x24, 0x14, 0x34, 0x0c, 0x2c, 0x1c, 0x3c,
61 0x02, 0x22, 0x12, 0x32, 0x0a, 0x2a, 0x1a, 0x3a,
62 0x06, 0x26, 0x16, 0x36, 0x0e, 0x2e, 0x1e, 0x3e,
63 0x01, 0x21, 0x11, 0x31, 0x09, 0x29, 0x19, 0x39,
64 0x05, 0x25, 0x15, 0x35, 0x0d, 0x2d, 0x1d, 0x3d,
65 0x03, 0x23, 0x13, 0x33, 0x0b, 0x2b, 0x1b, 0x3b,
66 0x07, 0x27, 0x17, 0x37, 0x0f, 0x2f, 0x1f, 0x3f};
68 /* Windowing function for Modified DCT - Thank you acroread */
69 static float window[] = {
70 0.00014, 0.00024, 0.00037, 0.00051, 0.00067, 0.00086, 0.00107, 0.00130,
71 0.00157, 0.00187, 0.00220, 0.00256, 0.00297, 0.00341, 0.00390, 0.00443,
72 0.00501, 0.00564, 0.00632, 0.00706, 0.00785, 0.00871, 0.00962, 0.01061,
73 0.01166, 0.01279, 0.01399, 0.01526, 0.01662, 0.01806, 0.01959, 0.02121,
74 0.02292, 0.02472, 0.02662, 0.02863, 0.03073, 0.03294, 0.03527, 0.03770,
75 0.04025, 0.04292, 0.04571, 0.04862, 0.05165, 0.05481, 0.05810, 0.06153,
76 0.06508, 0.06878, 0.07261, 0.07658, 0.08069, 0.08495, 0.08935, 0.09389,
77 0.09859, 0.10343, 0.10842, 0.11356, 0.11885, 0.12429, 0.12988, 0.13563,
78 0.14152, 0.14757, 0.15376, 0.16011, 0.16661, 0.17325, 0.18005, 0.18699,
79 0.19407, 0.20130, 0.20867, 0.21618, 0.22382, 0.23161, 0.23952, 0.24757,
80 0.25574, 0.26404, 0.27246, 0.28100, 0.28965, 0.29841, 0.30729, 0.31626,
81 0.32533, 0.33450, 0.34376, 0.35311, 0.36253, 0.37204, 0.38161, 0.39126,
82 0.40096, 0.41072, 0.42054, 0.43040, 0.44030, 0.45023, 0.46020, 0.47019,
83 0.48020, 0.49022, 0.50025, 0.51028, 0.52031, 0.53033, 0.54033, 0.55031,
84 0.56026, 0.57019, 0.58007, 0.58991, 0.59970, 0.60944, 0.61912, 0.62873,
85 0.63827, 0.64774, 0.65713, 0.66643, 0.67564, 0.68476, 0.69377, 0.70269,
86 0.71150, 0.72019, 0.72877, 0.73723, 0.74557, 0.75378, 0.76186, 0.76981,
87 0.77762, 0.78530, 0.79283, 0.80022, 0.80747, 0.81457, 0.82151, 0.82831,
88 0.83496, 0.84145, 0.84779, 0.85398, 0.86001, 0.86588, 0.87160, 0.87716,
89 0.88257, 0.88782, 0.89291, 0.89785, 0.90264, 0.90728, 0.91176, 0.91610,
90 0.92028, 0.92432, 0.92822, 0.93197, 0.93558, 0.93906, 0.94240, 0.94560,
91 0.94867, 0.95162, 0.95444, 0.95713, 0.95971, 0.96217, 0.96451, 0.96674,
92 0.96887, 0.97089, 0.97281, 0.97463, 0.97635, 0.97799, 0.97953, 0.98099,
93 0.98236, 0.98366, 0.98488, 0.98602, 0.98710, 0.98811, 0.98905, 0.98994,
94 0.99076, 0.99153, 0.99225, 0.99291, 0.99353, 0.99411, 0.99464, 0.99513,
95 0.99558, 0.99600, 0.99639, 0.99674, 0.99706, 0.99736, 0.99763, 0.99788,
96 0.99811, 0.99831, 0.99850, 0.99867, 0.99882, 0.99895, 0.99908, 0.99919,
97 0.99929, 0.99938, 0.99946, 0.99953, 0.99959, 0.99965, 0.99969, 0.99974,
98 0.99978, 0.99981, 0.99984, 0.99986, 0.99988, 0.99990, 0.99992, 0.99993,
99 0.99994, 0.99995, 0.99996, 0.99997, 0.99998, 0.99998, 0.99998, 0.99999,
100 0.99999, 0.99999, 0.99999, 1.00000, 1.00000, 1.00000, 1.00000, 1.00000,
101 1.00000, 1.00000, 1.00000, 1.00000, 1.00000, 1.00000, 1.00000, 1.00000 };
103 static __inline__ void swap_cmplx(complex_t *a, complex_t *b)
112 static __inline__ complex_t cmplx_mult(complex_t a, complex_t b)
116 ret.real = a.real * b.real - a.imag * b.imag;
117 ret.imag = a.real * b.imag + a.imag * b.real;
122 void imdct_init(imdct_t * p_imdct)
125 complex_t angle_step;
126 complex_t current_angle;
128 /* Twiddle factors to turn IFFT into IMDCT */
129 for (i=0; i < N/4; i++) {
130 p_imdct->xcos1[i] = -cos(2 * M_PI * (8*i+1)/(8*N)) ;
131 p_imdct->xsin1[i] = -sin(2 * M_PI * (8*i+1)/(8*N)) ;
134 /* More twiddle factors to turn IFFT into IMDCT */
135 for (i=0; i < N/8; i++) {
136 p_imdct->xcos2[i] = -cos(2 * M_PI * (8*i+1)/(4*N)) ;
137 p_imdct->xsin2[i] = -sin(2 * M_PI * (8*i+1)/(4*N)) ;
140 /* Canonical twiddle factors for FFT */
141 p_imdct->w[0] = p_imdct->w_1;
142 p_imdct->w[1] = p_imdct->w_2;
143 p_imdct->w[2] = p_imdct->w_4;
144 p_imdct->w[3] = p_imdct->w_8;
145 p_imdct->w[4] = p_imdct->w_16;
146 p_imdct->w[5] = p_imdct->w_32;
147 p_imdct->w[6] = p_imdct->w_64;
149 for (i = 0; i < 7; i++) {
150 angle_step.real = cos(-2.0f * M_PI / (1 << (i+1)));
151 angle_step.imag = sin(-2.0f * M_PI / (1 << (i+1)));
153 current_angle.real = 1.0f;
154 current_angle.imag = 0.0f;
156 for (k = 0; k < 1 << i; k++) {
157 p_imdct->w[i][k] = current_angle;
158 current_angle = cmplx_mult(current_angle,angle_step);
163 void imdct (ac3dec_t * p_ac3dec, s16 * buffer)
165 int i, i_stream_done;
167 void (*do_imdct)(imdct_t * p_imdct, float x[],float y[], int id);
169 /* Test if dm in frequency is doable */
170 if ( !(doable = p_ac3dec->audblk.blksw[0]) )
171 do_imdct = imdct_do_512;
173 do_imdct = imdct_do_256;
175 /* Downmix in the frequency domain if all the channes use the same imdct */
176 for (i=0; i < p_ac3dec->bsi.nfchans; i++)
178 if ( doable != p_ac3dec->audblk.blksw[i] )
187 i_stream_done = downmix(p_ac3dec, p_ac3dec->coeffs.fbw[0], buffer);
188 do_imdct(&p_ac3dec->imdct,p_ac3dec->coeffs.fbw[0],p_ac3dec->samples.channel[0], 0);
189 do_imdct(&p_ac3dec->imdct, p_ac3dec->coeffs.fbw[1],p_ac3dec->samples.channel[1], 1);
191 for (i=0; i<p_ac3dec->bsi.nfchans;i++) {
192 if (p_ac3dec->audblk.blksw[i])
193 imdct_do_256(&p_ac3dec->imdct, p_ac3dec->coeffs.fbw[i],p_ac3dec->samples.channel[i], i);
195 imdct_do_512(&p_ac3dec->imdct, p_ac3dec->coeffs.fbw[i],p_ac3dec->samples.channel[i], i);
197 i_stream_done = downmix(p_ac3dec, p_ac3dec->samples.channel[0], buffer);
200 if ( !i_stream_done ) /* We have to stream sample */
202 stream_sample_2ch_to_s16_c(buffer, p_ac3dec->samples.channel[0],
203 p_ac3dec->samples.channel[1]);
205 stream_sample_1ch_to_s16_c(buffer, p_ac3dec->samples.channel[0]);
208 /* XXX?? We don't bother with the IMDCT for the LFE as it's currently
211 // imdct_do_512(coeffs->lfe,samples->channel[5],delay[5]);
214 void imdct_do_512(imdct_t * p_imdct, float x[], float y[], int id)
232 /* Pre IFFT complex multiply plus IFFT cmplx conjugate */
233 for (i=0; i < N/4; i++) {
234 /* z[i] = (X[N/2-2*i-1] + j * X[2*i]) * (xcos1[i] + j * xsin1[i]) ; */
235 p_imdct->buf[i].real = (x[N/2-2*i-1] * p_imdct->xcos1[i]) - (x[2*i] * p_imdct->xsin1[i]);
236 p_imdct->buf[i].imag = -((x[2*i] * p_imdct->xcos1[i]) + (x[N/2-2*i-1] * p_imdct->xsin1[i]));
239 /* Bit reversed shuffling */
240 for (i=0; i<N/4; i++) {
241 k = bit_reverse_512[i];
243 swap_cmplx(&p_imdct->buf[i],&p_imdct->buf[k]);
247 for (m=0; m < 7; m++) {
249 two_m_plus_one = (1 << (m+1));
251 for (k = 0; k < two_m; k++) {
252 for (i = 0; i < 128; i += two_m_plus_one) {
255 tmp_a_r = p_imdct->buf[p].real;
256 tmp_a_i = p_imdct->buf[p].imag;
257 tmp_b_r = p_imdct->buf[q].real * p_imdct->w[m][k].real - p_imdct->buf[q].imag * p_imdct->w[m][k].imag;
258 tmp_b_i = p_imdct->buf[q].imag * p_imdct->w[m][k].real + p_imdct->buf[q].real * p_imdct->w[m][k].imag;
259 p_imdct->buf[p].real = tmp_a_r + tmp_b_r;
260 p_imdct->buf[p].imag = tmp_a_i + tmp_b_i;
261 p_imdct->buf[q].real = tmp_a_r - tmp_b_r;
262 p_imdct->buf[q].imag = tmp_a_i - tmp_b_i;
267 /* Post IFFT complex multiply plus IFFT complex conjugate*/
268 for (i=0; i < N/4; i++) {
269 /* y[n] = z[n] * (xcos1[n] + j * xsin1[n]) ; */
270 tmp_a_r = p_imdct->buf[i].real;
271 tmp_a_i = - p_imdct->buf[i].imag;
272 p_imdct->buf[i].real =(tmp_a_r * p_imdct->xcos1[i]) - (tmp_a_i * p_imdct->xsin1[i]);
273 p_imdct->buf[i].imag =(tmp_a_r * p_imdct->xsin1[i]) + (tmp_a_i * p_imdct->xcos1[i]);
277 delay_ptr = p_imdct->delay[id];
279 /* Window and convert to real valued signal */
280 for (i=0; i<N/8; i++) {
281 *y_ptr++ = 2.0f * (-p_imdct->buf[N/8+i].imag * *window_ptr++ + *delay_ptr++);
282 *y_ptr++ = 2.0f * (p_imdct->buf[N/8-i-1].real * *window_ptr++ + *delay_ptr++);
285 for (i=0; i<N/8; i++) {
286 *y_ptr++ = 2.0f * (-p_imdct->buf[i].real * *window_ptr++ + *delay_ptr++);
287 *y_ptr++ = 2.0f * (p_imdct->buf[N/4-i-1].imag * *window_ptr++ + *delay_ptr++);
290 /* The trailing edge of the window goes into the delay line */
291 delay_ptr = p_imdct->delay[id];
293 for (i=0; i<N/8; i++) {
294 *delay_ptr++ = -p_imdct->buf[N/8+i].real * *--window_ptr;
295 *delay_ptr++ = p_imdct->buf[N/8-i-1].imag * *--window_ptr;
298 for (i=0; i<N/8; i++) {
299 *delay_ptr++ = p_imdct->buf[i].imag * *--window_ptr;
300 *delay_ptr++ = -p_imdct->buf[N/4-i-1].real * *--window_ptr;
304 void imdct_do_256(imdct_t * p_imdct,float x[],float y[], int id)
317 complex_t *buf_1, *buf_2;
319 buf_1 = &p_imdct->buf[0];
320 buf_2 = &p_imdct->buf[64];
322 /* Pre IFFT complex multiply plus IFFT cmplx conjugate */
323 for (k=0; k<N/8; k++) {
325 /* X2[k] = X[2*k+1] */
330 /* Z1[k] = (X1[N/4-2*k-1] + j * X1[2*k]) * (xcos2[k] + j * xsin2[k]); */
331 buf_1[k].real = x[p] * p_imdct->xcos2[k] - x[q] * p_imdct->xsin2[k];
332 buf_1[k].imag = - (x[q] * p_imdct->xcos2[k] + x[p] * p_imdct->xsin2[k]);
333 /* Z2[k] = (X2[N/4-2*k-1] + j * X2[2*k]) * (xcos2[k] + j * xsin2[k]); */
334 buf_2[k].real = x[p + 1] * p_imdct->xcos2[k] - x[q + 1] * p_imdct->xsin2[k];
335 buf_2[k].imag = - (x[q + 1] * p_imdct->xcos2[k] + x[p + 1] * p_imdct->xsin2[k]);
338 /* IFFT Bit reversed shuffling */
339 for (i=0; i<N/8; i++) {
340 k = bit_reverse_256[i];
342 swap_cmplx(&buf_1[i],&buf_1[k]);
343 swap_cmplx(&buf_2[i],&buf_2[k]);
348 for (m=0; m < 6; m++) {
350 two_m_plus_one = (1 << (m+1));
352 for (k = 0; k < two_m; k++) {
353 for (i = 0; i < 64; i += two_m_plus_one) {
357 tmp_a_r = buf_1[p].real;
358 tmp_a_i = buf_1[p].imag;
359 tmp_b_r = buf_1[q].real * p_imdct->w[m][k].real - buf_1[q].imag * p_imdct->w[m][k].imag;
360 tmp_b_i = buf_1[q].imag * p_imdct->w[m][k].real + buf_1[q].real * p_imdct->w[m][k].imag;
361 buf_1[p].real = tmp_a_r + tmp_b_r;
362 buf_1[p].imag = tmp_a_i + tmp_b_i;
363 buf_1[q].real = tmp_a_r - tmp_b_r;
364 buf_1[q].imag = tmp_a_i - tmp_b_i;
367 tmp_a_r = buf_2[p].real;
368 tmp_a_i = buf_2[p].imag;
369 tmp_b_r = buf_2[q].real * p_imdct->w[m][k].real - buf_2[q].imag * p_imdct->w[m][k].imag;
370 tmp_b_i = buf_2[q].imag * p_imdct->w[m][k].real + buf_2[q].real * p_imdct->w[m][k].imag;
371 buf_2[p].real = tmp_a_r + tmp_b_r;
372 buf_2[p].imag = tmp_a_i + tmp_b_i;
373 buf_2[q].real = tmp_a_r - tmp_b_r;
374 buf_2[q].imag = tmp_a_i - tmp_b_i;
379 /* Post IFFT complex multiply */
380 for (i=0; i < N/8; i++) {
381 /* y1[n] = z1[n] * (xcos2[n] + j * xs in2[n]) ; */
382 tmp_a_r = buf_1[i].real;
383 tmp_a_i = - buf_1[i].imag;
384 buf_1[i].real =(tmp_a_r * p_imdct->xcos2[i]) - (tmp_a_i * p_imdct->xsin2[i]);
385 buf_1[i].imag =(tmp_a_r * p_imdct->xsin2[i]) + (tmp_a_i * p_imdct->xcos2[i]);
386 /* y2[n] = z2[n] * (xcos2[n] + j * xsin2[n]) ; */
387 tmp_a_r = buf_2[i].real;
388 tmp_a_i = - buf_2[i].imag;
389 buf_2[i].real =(tmp_a_r * p_imdct->xcos2[i]) - (tmp_a_i * p_imdct->xsin2[i]);
390 buf_2[i].imag =(tmp_a_r * p_imdct->xsin2[i]) + (tmp_a_i * p_imdct->xcos2[i]);
393 /* Window and convert to real valued signal */
394 for (i=0; i<N/8; i++) {
395 y[2*i] = -buf_1[i].imag * window[2*i];
396 y[2*i+1] = buf_1[N/8-i-1].real * window[2*i+1];
397 y[N/4+2*i] = -buf_1[i].real * window[N/4+2*i];
398 y[N/4+2*i+1] = buf_1[N/8-i-1].imag * window[N/4+2*i+1];
399 y[N/2+2*i] = -buf_2[i].real * window[N/2-2*i-1];
400 y[N/2+2*i+1] = buf_2[N/8-i-1].imag * window[N/2-2*i-2];
401 y[3*N/4+2*i] = buf_2[i].imag * window[N/4-2*i-1];
402 y[3*N/4+2*i+1] = -buf_2[N/8-i-1].real * window[N/4-2*i-2];
405 /* Overlap and add */
406 for (i=0; i<N/2; i++) {
407 y[i] = 2 * (y[i] + p_imdct->delay[id][i]);
408 p_imdct->delay[id][i] = y[N/2+i];