Line | Branch | Exec | Source |
---|---|---|---|
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 "libavutil/attributes.h" | ||
27 | #include "libavutil/intfloat.h" | ||
28 | #include "sbrdsp.h" | ||
29 | |||
30 | 251188 | static float sbr_sum_square_c(float (*x)[2], int n) | |
31 | { | ||
32 | 251188 | float sum0 = 0.0f, sum1 = 0.0f; | |
33 | int i; | ||
34 | |||
35 |
2/2✓ Branch 0 taken 2752464 times.
✓ Branch 1 taken 251188 times.
|
3003652 | for (i = 0; i < n; i += 2) |
36 | { | ||
37 | 2752464 | sum0 += x[i + 0][0] * x[i + 0][0]; | |
38 | 2752464 | sum1 += x[i + 0][1] * x[i + 0][1]; | |
39 | 2752464 | sum0 += x[i + 1][0] * x[i + 1][0]; | |
40 | 2752464 | sum1 += x[i + 1][1] * x[i + 1][1]; | |
41 | } | ||
42 | |||
43 | 251188 | return sum0 + sum1; | |
44 | } | ||
45 | |||
46 | 355459 | static void sbr_neg_odd_64_c(float *x) | |
47 | { | ||
48 | 355459 | union av_intfloat32 *xi = (union av_intfloat32*) x; | |
49 | int i; | ||
50 |
2/2✓ Branch 0 taken 5687344 times.
✓ Branch 1 taken 355459 times.
|
6042803 | for (i = 1; i < 64; i += 4) { |
51 | 5687344 | xi[i + 0].i ^= 1U << 31; | |
52 | 5687344 | xi[i + 2].i ^= 1U << 31; | |
53 | } | ||
54 | 355459 | } | |
55 | |||
56 | 365091 | static void sbr_qmf_pre_shuffle_c(float *z) | |
57 | { | ||
58 | 365091 | union av_intfloat32 *zi = (union av_intfloat32*) z; | |
59 | int k; | ||
60 | 365091 | zi[64].i = zi[0].i; | |
61 | 365091 | zi[65].i = zi[1].i; | |
62 |
2/2✓ Branch 0 taken 5476365 times.
✓ Branch 1 taken 365091 times.
|
5841456 | for (k = 1; k < 31; k += 2) { |
63 | 5476365 | zi[64 + 2 * k + 0].i = zi[64 - k].i ^ (1U << 31); | |
64 | 5476365 | zi[64 + 2 * k + 1].i = zi[ k + 1].i; | |
65 | 5476365 | zi[64 + 2 * k + 2].i = zi[63 - k].i ^ (1U << 31); | |
66 | 5476365 | zi[64 + 2 * k + 3].i = zi[ k + 2].i; | |
67 | } | ||
68 | |||
69 | 365091 | zi[64 + 2 * 31 + 0].i = zi[64 - 31].i ^ (1U << 31); | |
70 | 365091 | zi[64 + 2 * 31 + 1].i = zi[31 + 1].i; | |
71 | 365091 | } | |
72 | |||
73 | 365091 | static void sbr_qmf_post_shuffle_c(float W[32][2], const float *z) | |
74 | { | ||
75 | 365091 | const union av_intfloat32 *zi = (const union av_intfloat32*) z; | |
76 | 365091 | union av_intfloat32 *Wi = (union av_intfloat32*) W; | |
77 | int k; | ||
78 |
2/2✓ Branch 0 taken 5841456 times.
✓ Branch 1 taken 365091 times.
|
6206547 | for (k = 0; k < 32; k += 2) { |
79 | 5841456 | Wi[2 * k + 0].i = zi[63 - k].i ^ (1U << 31); | |
80 | 5841456 | Wi[2 * k + 1].i = zi[ k + 0].i; | |
81 | 5841456 | Wi[2 * k + 2].i = zi[62 - k].i ^ (1U << 31); | |
82 | 5841456 | Wi[2 * k + 3].i = zi[ k + 1].i; | |
83 | } | ||
84 | 365091 | } | |
85 | |||
86 | 73603 | static void sbr_qmf_deint_neg_c(float *v, const float *src) | |
87 | { | ||
88 | 73603 | const union av_intfloat32 *si = (const union av_intfloat32*)src; | |
89 | 73603 | union av_intfloat32 *vi = (union av_intfloat32*)v; | |
90 | int i; | ||
91 |
2/2✓ Branch 0 taken 2355296 times.
✓ Branch 1 taken 73603 times.
|
2428899 | for (i = 0; i < 32; i++) { |
92 | 2355296 | vi[ i].i = si[63 - 2 * i ].i; | |
93 | 2355296 | vi[63 - i].i = si[63 - 2 * i - 1].i ^ (1U << 31); | |
94 | } | ||
95 | 73603 | } | |
96 | |||
97 | #if 0 | ||
98 | /* This code is slower because it multiplies memory accesses. | ||
99 | * It is left for educational purposes and because it may offer | ||
100 | * a better reference for writing arch-specific DSP functions. */ | ||
101 | static av_always_inline void autocorrelate(const float x[40][2], | ||
102 | float phi[3][2][2], int lag) | ||
103 | { | ||
104 | int i; | ||
105 | float real_sum = 0.0f; | ||
106 | float imag_sum = 0.0f; | ||
107 | if (lag) { | ||
108 | for (i = 1; i < 38; i++) { | ||
109 | real_sum += x[i][0] * x[i+lag][0] + x[i][1] * x[i+lag][1]; | ||
110 | imag_sum += x[i][0] * x[i+lag][1] - x[i][1] * x[i+lag][0]; | ||
111 | } | ||
112 | phi[2-lag][1][0] = real_sum + x[ 0][0] * x[lag][0] + x[ 0][1] * x[lag][1]; | ||
113 | phi[2-lag][1][1] = imag_sum + x[ 0][0] * x[lag][1] - x[ 0][1] * x[lag][0]; | ||
114 | if (lag == 1) { | ||
115 | phi[0][0][0] = real_sum + x[38][0] * x[39][0] + x[38][1] * x[39][1]; | ||
116 | phi[0][0][1] = imag_sum + x[38][0] * x[39][1] - x[38][1] * x[39][0]; | ||
117 | } | ||
118 | } else { | ||
119 | for (i = 1; i < 38; i++) { | ||
120 | real_sum += x[i][0] * x[i][0] + x[i][1] * x[i][1]; | ||
121 | } | ||
122 | phi[2][1][0] = real_sum + x[ 0][0] * x[ 0][0] + x[ 0][1] * x[ 0][1]; | ||
123 | phi[1][0][0] = real_sum + x[38][0] * x[38][0] + x[38][1] * x[38][1]; | ||
124 | } | ||
125 | } | ||
126 | |||
127 | static void sbr_autocorrelate_c(const float x[40][2], float phi[3][2][2]) | ||
128 | { | ||
129 | autocorrelate(x, phi, 0); | ||
130 | autocorrelate(x, phi, 1); | ||
131 | autocorrelate(x, phi, 2); | ||
132 | } | ||
133 | #else | ||
134 | 124327 | static void sbr_autocorrelate_c(const float x[40][2], float phi[3][2][2]) | |
135 | { | ||
136 | 124327 | float real_sum2 = x[0][0] * x[2][0] + x[0][1] * x[2][1]; | |
137 | 124327 | float imag_sum2 = x[0][0] * x[2][1] - x[0][1] * x[2][0]; | |
138 | 124327 | float real_sum1 = 0.0f, imag_sum1 = 0.0f, real_sum0 = 0.0f; | |
139 | int i; | ||
140 |
2/2✓ Branch 0 taken 4600099 times.
✓ Branch 1 taken 124327 times.
|
4724426 | for (i = 1; i < 38; i++) { |
141 | 4600099 | real_sum0 += x[i][0] * x[i ][0] + x[i][1] * x[i ][1]; | |
142 | 4600099 | real_sum1 += x[i][0] * x[i + 1][0] + x[i][1] * x[i + 1][1]; | |
143 | 4600099 | imag_sum1 += x[i][0] * x[i + 1][1] - x[i][1] * x[i + 1][0]; | |
144 | 4600099 | real_sum2 += x[i][0] * x[i + 2][0] + x[i][1] * x[i + 2][1]; | |
145 | 4600099 | imag_sum2 += x[i][0] * x[i + 2][1] - x[i][1] * x[i + 2][0]; | |
146 | } | ||
147 | 124327 | phi[2 - 2][1][0] = real_sum2; | |
148 | 124327 | phi[2 - 2][1][1] = imag_sum2; | |
149 | 124327 | phi[2 ][1][0] = real_sum0 + x[ 0][0] * x[ 0][0] + x[ 0][1] * x[ 0][1]; | |
150 | 124327 | phi[1 ][0][0] = real_sum0 + x[38][0] * x[38][0] + x[38][1] * x[38][1]; | |
151 | 124327 | phi[2 - 1][1][0] = real_sum1 + x[ 0][0] * x[ 1][0] + x[ 0][1] * x[ 1][1]; | |
152 | 124327 | phi[2 - 1][1][1] = imag_sum1 + x[ 0][0] * x[ 1][1] - x[ 0][1] * x[ 1][0]; | |
153 | 124327 | phi[0 ][0][0] = real_sum1 + x[38][0] * x[39][0] + x[38][1] * x[39][1]; | |
154 | 124327 | phi[0 ][0][1] = imag_sum1 + x[38][0] * x[39][1] - x[38][1] * x[39][0]; | |
155 | 124327 | } | |
156 | #endif | ||
157 | |||
158 | 172098 | static void sbr_hf_gen_c(float (*X_high)[2], const float (*X_low)[2], | |
159 | const float alpha0[2], const float alpha1[2], | ||
160 | float bw, int start, int end) | ||
161 | { | ||
162 | float alpha[4]; | ||
163 | int i; | ||
164 | |||
165 | 172098 | alpha[0] = alpha1[0] * bw * bw; | |
166 | 172098 | alpha[1] = alpha1[1] * bw * bw; | |
167 | 172098 | alpha[2] = alpha0[0] * bw; | |
168 | 172098 | alpha[3] = alpha0[1] * bw; | |
169 | |||
170 |
2/2✓ Branch 0 taken 5513088 times.
✓ Branch 1 taken 172098 times.
|
5685186 | for (i = start; i < end; i++) { |
171 | 5513088 | X_high[i][0] = | |
172 | 5513088 | X_low[i - 2][0] * alpha[0] - | |
173 | 5513088 | X_low[i - 2][1] * alpha[1] + | |
174 | 5513088 | X_low[i - 1][0] * alpha[2] - | |
175 | 5513088 | X_low[i - 1][1] * alpha[3] + | |
176 | 5513088 | X_low[i][0]; | |
177 | 5513088 | X_high[i][1] = | |
178 | 5513088 | X_low[i - 2][1] * alpha[0] + | |
179 | 5513088 | X_low[i - 2][0] * alpha[1] + | |
180 | 5513088 | X_low[i - 1][1] * alpha[2] + | |
181 | 5513088 | X_low[i - 1][0] * alpha[3] + | |
182 | 5513088 | X_low[i][1]; | |
183 | } | ||
184 | 172098 | } | |
185 | |||
186 | 212195 | static void sbr_hf_g_filt_c(float (*Y)[2], const float (*X_high)[40][2], | |
187 | const float *g_filt, int m_max, intptr_t ixh) | ||
188 | { | ||
189 | int m; | ||
190 | |||
191 |
2/2✓ Branch 0 taken 5504544 times.
✓ Branch 1 taken 212195 times.
|
5716739 | for (m = 0; m < m_max; m++) { |
192 | 5504544 | Y[m][0] = X_high[m][ixh][0] * g_filt[m]; | |
193 | 5504544 | Y[m][1] = X_high[m][ixh][1] * g_filt[m]; | |
194 | } | ||
195 | 212195 | } | |
196 | |||
197 | 209994 | static av_always_inline void sbr_hf_apply_noise(float (*Y)[2], | |
198 | const float *s_m, | ||
199 | const float *q_filt, | ||
200 | int noise, | ||
201 | float phi_sign0, | ||
202 | float phi_sign1, | ||
203 | int m_max) | ||
204 | { | ||
205 | int m; | ||
206 | |||
207 |
2/2✓ Branch 0 taken 5449888 times.
✓ Branch 1 taken 209994 times.
|
5659882 | for (m = 0; m < m_max; m++) { |
208 | 5449888 | float y0 = Y[m][0]; | |
209 | 5449888 | float y1 = Y[m][1]; | |
210 | 5449888 | noise = (noise + 1) & 0x1ff; | |
211 |
2/2✓ Branch 0 taken 37758 times.
✓ Branch 1 taken 5412130 times.
|
5449888 | if (s_m[m]) { |
212 | 37758 | y0 += s_m[m] * phi_sign0; | |
213 | 37758 | y1 += s_m[m] * phi_sign1; | |
214 | } else { | ||
215 | 5412130 | y0 += q_filt[m] * ff_sbr_noise_table[noise][0]; | |
216 | 5412130 | y1 += q_filt[m] * ff_sbr_noise_table[noise][1]; | |
217 | } | ||
218 | 5449888 | Y[m][0] = y0; | |
219 | 5449888 | Y[m][1] = y1; | |
220 | 5449888 | phi_sign1 = -phi_sign1; | |
221 | } | ||
222 | 209994 | } | |
223 | |||
224 | #include "sbrdsp_template.c" | ||
225 |