xref: /third_party/ffmpeg/libavcodec/sbrdsp.c (revision cabdff1a)
1/*
2 * AAC Spectral Band Replication decoding functions
3 * Copyright (c) 2008-2009 Robert Swain ( rob opendot cl )
4 * Copyright (c) 2009-2010 Alex Converse <alex.converse@gmail.com>
5 *
6 * This file is part of FFmpeg.
7 *
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
12 *
13 * FFmpeg is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
16 * Lesser General Public License for more details.
17 *
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 */
22
23#define USE_FIXED 0
24
25#include "aac.h"
26#include "config.h"
27#include "libavutil/attributes.h"
28#include "libavutil/intfloat.h"
29#include "sbrdsp.h"
30
31static float sbr_sum_square_c(float (*x)[2], int n)
32{
33    float sum0 = 0.0f, sum1 = 0.0f;
34    int i;
35
36    for (i = 0; i < n; i += 2)
37    {
38        sum0 += x[i + 0][0] * x[i + 0][0];
39        sum1 += x[i + 0][1] * x[i + 0][1];
40        sum0 += x[i + 1][0] * x[i + 1][0];
41        sum1 += x[i + 1][1] * x[i + 1][1];
42    }
43
44    return sum0 + sum1;
45}
46
47static void sbr_neg_odd_64_c(float *x)
48{
49    union av_intfloat32 *xi = (union av_intfloat32*) x;
50    int i;
51    for (i = 1; i < 64; i += 4) {
52        xi[i + 0].i ^= 1U << 31;
53        xi[i + 2].i ^= 1U << 31;
54    }
55}
56
57static void sbr_qmf_pre_shuffle_c(float *z)
58{
59    union av_intfloat32 *zi = (union av_intfloat32*) z;
60    int k;
61    zi[64].i = zi[0].i;
62    zi[65].i = zi[1].i;
63    for (k = 1; k < 31; k += 2) {
64        zi[64 + 2 * k + 0].i = zi[64 - k].i ^ (1U << 31);
65        zi[64 + 2 * k + 1].i = zi[ k + 1].i;
66        zi[64 + 2 * k + 2].i = zi[63 - k].i ^ (1U << 31);
67        zi[64 + 2 * k + 3].i = zi[ k + 2].i;
68    }
69
70    zi[64 + 2 * 31 + 0].i = zi[64 - 31].i ^ (1U << 31);
71    zi[64 + 2 * 31 + 1].i = zi[31 +  1].i;
72}
73
74static void sbr_qmf_post_shuffle_c(float W[32][2], const float *z)
75{
76    const union av_intfloat32 *zi = (const union av_intfloat32*) z;
77    union av_intfloat32 *Wi       = (union av_intfloat32*) W;
78    int k;
79    for (k = 0; k < 32; k += 2) {
80        Wi[2 * k + 0].i = zi[63 - k].i ^ (1U << 31);
81        Wi[2 * k + 1].i = zi[ k + 0].i;
82        Wi[2 * k + 2].i = zi[62 - k].i ^ (1U << 31);
83        Wi[2 * k + 3].i = zi[ k + 1].i;
84    }
85}
86
87static void sbr_qmf_deint_neg_c(float *v, const float *src)
88{
89    const union av_intfloat32 *si = (const union av_intfloat32*)src;
90    union av_intfloat32 *vi = (union av_intfloat32*)v;
91    int i;
92    for (i = 0; i < 32; i++) {
93        vi[     i].i = si[63 - 2 * i    ].i;
94        vi[63 - i].i = si[63 - 2 * i - 1].i ^ (1U << 31);
95    }
96}
97
98#if 0
99    /* This code is slower because it multiplies memory accesses.
100     * It is left for educational purposes and because it may offer
101     * a better reference for writing arch-specific DSP functions. */
102static av_always_inline void autocorrelate(const float x[40][2],
103                                           float phi[3][2][2], int lag)
104{
105    int i;
106    float real_sum = 0.0f;
107    float imag_sum = 0.0f;
108    if (lag) {
109        for (i = 1; i < 38; i++) {
110            real_sum += x[i][0] * x[i+lag][0] + x[i][1] * x[i+lag][1];
111            imag_sum += x[i][0] * x[i+lag][1] - x[i][1] * x[i+lag][0];
112        }
113        phi[2-lag][1][0] = real_sum + x[ 0][0] * x[lag][0] + x[ 0][1] * x[lag][1];
114        phi[2-lag][1][1] = imag_sum + x[ 0][0] * x[lag][1] - x[ 0][1] * x[lag][0];
115        if (lag == 1) {
116            phi[0][0][0] = real_sum + x[38][0] * x[39][0] + x[38][1] * x[39][1];
117            phi[0][0][1] = imag_sum + x[38][0] * x[39][1] - x[38][1] * x[39][0];
118        }
119    } else {
120        for (i = 1; i < 38; i++) {
121            real_sum += x[i][0] * x[i][0] + x[i][1] * x[i][1];
122        }
123        phi[2][1][0] = real_sum + x[ 0][0] * x[ 0][0] + x[ 0][1] * x[ 0][1];
124        phi[1][0][0] = real_sum + x[38][0] * x[38][0] + x[38][1] * x[38][1];
125    }
126}
127
128static void sbr_autocorrelate_c(const float x[40][2], float phi[3][2][2])
129{
130    autocorrelate(x, phi, 0);
131    autocorrelate(x, phi, 1);
132    autocorrelate(x, phi, 2);
133}
134#else
135static void sbr_autocorrelate_c(const float x[40][2], float phi[3][2][2])
136{
137    float real_sum2 = x[0][0] * x[2][0] + x[0][1] * x[2][1];
138    float imag_sum2 = x[0][0] * x[2][1] - x[0][1] * x[2][0];
139    float real_sum1 = 0.0f, imag_sum1 = 0.0f, real_sum0 = 0.0f;
140    int   i;
141    for (i = 1; i < 38; i++) {
142        real_sum0 += x[i][0] * x[i    ][0] + x[i][1] * x[i    ][1];
143        real_sum1 += x[i][0] * x[i + 1][0] + x[i][1] * x[i + 1][1];
144        imag_sum1 += x[i][0] * x[i + 1][1] - x[i][1] * x[i + 1][0];
145        real_sum2 += x[i][0] * x[i + 2][0] + x[i][1] * x[i + 2][1];
146        imag_sum2 += x[i][0] * x[i + 2][1] - x[i][1] * x[i + 2][0];
147    }
148    phi[2 - 2][1][0] = real_sum2;
149    phi[2 - 2][1][1] = imag_sum2;
150    phi[2    ][1][0] = real_sum0 + x[ 0][0] * x[ 0][0] + x[ 0][1] * x[ 0][1];
151    phi[1    ][0][0] = real_sum0 + x[38][0] * x[38][0] + x[38][1] * x[38][1];
152    phi[2 - 1][1][0] = real_sum1 + x[ 0][0] * x[ 1][0] + x[ 0][1] * x[ 1][1];
153    phi[2 - 1][1][1] = imag_sum1 + x[ 0][0] * x[ 1][1] - x[ 0][1] * x[ 1][0];
154    phi[0    ][0][0] = real_sum1 + x[38][0] * x[39][0] + x[38][1] * x[39][1];
155    phi[0    ][0][1] = imag_sum1 + x[38][0] * x[39][1] - x[38][1] * x[39][0];
156}
157#endif
158
159static void sbr_hf_gen_c(float (*X_high)[2], const float (*X_low)[2],
160                         const float alpha0[2], const float alpha1[2],
161                         float bw, int start, int end)
162{
163    float alpha[4];
164    int i;
165
166    alpha[0] = alpha1[0] * bw * bw;
167    alpha[1] = alpha1[1] * bw * bw;
168    alpha[2] = alpha0[0] * bw;
169    alpha[3] = alpha0[1] * bw;
170
171    for (i = start; i < end; i++) {
172        X_high[i][0] =
173            X_low[i - 2][0] * alpha[0] -
174            X_low[i - 2][1] * alpha[1] +
175            X_low[i - 1][0] * alpha[2] -
176            X_low[i - 1][1] * alpha[3] +
177            X_low[i][0];
178        X_high[i][1] =
179            X_low[i - 2][1] * alpha[0] +
180            X_low[i - 2][0] * alpha[1] +
181            X_low[i - 1][1] * alpha[2] +
182            X_low[i - 1][0] * alpha[3] +
183            X_low[i][1];
184    }
185}
186
187static void sbr_hf_g_filt_c(float (*Y)[2], const float (*X_high)[40][2],
188                            const float *g_filt, int m_max, intptr_t ixh)
189{
190    int m;
191
192    for (m = 0; m < m_max; m++) {
193        Y[m][0] = X_high[m][ixh][0] * g_filt[m];
194        Y[m][1] = X_high[m][ixh][1] * g_filt[m];
195    }
196}
197
198static av_always_inline void sbr_hf_apply_noise(float (*Y)[2],
199                                                const float *s_m,
200                                                const float *q_filt,
201                                                int noise,
202                                                float phi_sign0,
203                                                float phi_sign1,
204                                                int m_max)
205{
206    int m;
207
208    for (m = 0; m < m_max; m++) {
209        float y0 = Y[m][0];
210        float y1 = Y[m][1];
211        noise = (noise + 1) & 0x1ff;
212        if (s_m[m]) {
213            y0 += s_m[m] * phi_sign0;
214            y1 += s_m[m] * phi_sign1;
215        } else {
216            y0 += q_filt[m] * ff_sbr_noise_table[noise][0];
217            y1 += q_filt[m] * ff_sbr_noise_table[noise][1];
218        }
219        Y[m][0] = y0;
220        Y[m][1] = y1;
221        phi_sign1 = -phi_sign1;
222    }
223}
224
225#include "sbrdsp_template.c"
226