1/*
2 * MDCT/IMDCT transforms
3 * Copyright (c) 2002 Fabrice Bellard
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22#include <stdlib.h>
23#include <string.h>
24#include "libavutil/common.h"
25#include "libavutil/mathematics.h"
26#include "fft.h"
27#include "fft-internal.h"
28
29/**
30 * @file
31 * MDCT/IMDCT transforms.
32 */
33
34#if FFT_FLOAT
35#   define RSCALE(x) (x)
36#else
37#if FFT_FIXED_32
38#   define RSCALE(x) (((x) + 32) >> 6)
39#else /* FFT_FIXED_32 */
40#   define RSCALE(x) ((x) >> 1)
41#endif /* FFT_FIXED_32 */
42#endif
43
44/**
45 * init MDCT or IMDCT computation.
46 */
47av_cold int ff_mdct_init(FFTContext *s, int nbits, int inverse, double scale)
48{
49    int n, n4, i;
50    double alpha, theta;
51    int tstep;
52
53    memset(s, 0, sizeof(*s));
54    n = 1 << nbits;
55    s->mdct_bits = nbits;
56    s->mdct_size = n;
57    n4 = n >> 2;
58    s->mdct_permutation = FF_MDCT_PERM_NONE;
59
60    if (ff_fft_init(s, s->mdct_bits - 2, inverse) < 0)
61        goto fail;
62
63    s->tcos = av_malloc_array(n/2, sizeof(FFTSample));
64    if (!s->tcos)
65        goto fail;
66
67    switch (s->mdct_permutation) {
68    case FF_MDCT_PERM_NONE:
69        s->tsin = s->tcos + n4;
70        tstep = 1;
71        break;
72    case FF_MDCT_PERM_INTERLEAVE:
73        s->tsin = s->tcos + 1;
74        tstep = 2;
75        break;
76    default:
77        goto fail;
78    }
79
80    theta = 1.0 / 8.0 + (scale < 0 ? n4 : 0);
81    scale = sqrt(fabs(scale));
82    for(i=0;i<n4;i++) {
83        alpha = 2 * M_PI * (i + theta) / n;
84        s->tcos[i*tstep] = FIX15(-cos(alpha) * scale);
85        s->tsin[i*tstep] = FIX15(-sin(alpha) * scale);
86    }
87    return 0;
88 fail:
89    ff_mdct_end(s);
90    return -1;
91}
92
93/**
94 * Compute the middle half of the inverse MDCT of size N = 2^nbits,
95 * thus excluding the parts that can be derived by symmetry
96 * @param output N/2 samples
97 * @param input N/2 samples
98 */
99void ff_imdct_half_c(FFTContext *s, FFTSample *output, const FFTSample *input)
100{
101    int k, n8, n4, n2, n, j;
102    const uint16_t *revtab = s->revtab;
103    const FFTSample *tcos = s->tcos;
104    const FFTSample *tsin = s->tsin;
105    const FFTSample *in1, *in2;
106    FFTComplex *z = (FFTComplex *)output;
107
108    n = 1 << s->mdct_bits;
109    n2 = n >> 1;
110    n4 = n >> 2;
111    n8 = n >> 3;
112
113    /* pre rotation */
114    in1 = input;
115    in2 = input + n2 - 1;
116    for(k = 0; k < n4; k++) {
117        j=revtab[k];
118        CMUL(z[j].re, z[j].im, *in2, *in1, tcos[k], tsin[k]);
119        in1 += 2;
120        in2 -= 2;
121    }
122    s->fft_calc(s, z);
123
124    /* post rotation + reordering */
125    for(k = 0; k < n8; k++) {
126        FFTSample r0, i0, r1, i1;
127        CMUL(r0, i1, z[n8-k-1].im, z[n8-k-1].re, tsin[n8-k-1], tcos[n8-k-1]);
128        CMUL(r1, i0, z[n8+k  ].im, z[n8+k  ].re, tsin[n8+k  ], tcos[n8+k  ]);
129        z[n8-k-1].re = r0;
130        z[n8-k-1].im = i0;
131        z[n8+k  ].re = r1;
132        z[n8+k  ].im = i1;
133    }
134}
135
136/**
137 * Compute inverse MDCT of size N = 2^nbits
138 * @param output N samples
139 * @param input N/2 samples
140 */
141void ff_imdct_calc_c(FFTContext *s, FFTSample *output, const FFTSample *input)
142{
143    int k;
144    int n = 1 << s->mdct_bits;
145    int n2 = n >> 1;
146    int n4 = n >> 2;
147
148    ff_imdct_half_c(s, output+n4, input);
149
150    for(k = 0; k < n4; k++) {
151        output[k] = -output[n2-k-1];
152        output[n-k-1] = output[n2+k];
153    }
154}
155
156/**
157 * Compute MDCT of size N = 2^nbits
158 * @param input N samples
159 * @param out N/2 samples
160 */
161void ff_mdct_calc_c(FFTContext *s, FFTSample *out, const FFTSample *input)
162{
163    int i, j, n, n8, n4, n2, n3;
164    FFTDouble re, im;
165    const uint16_t *revtab = s->revtab;
166    const FFTSample *tcos = s->tcos;
167    const FFTSample *tsin = s->tsin;
168    FFTComplex *x = (FFTComplex *)out;
169
170    n = 1 << s->mdct_bits;
171    n2 = n >> 1;
172    n4 = n >> 2;
173    n8 = n >> 3;
174    n3 = 3 * n4;
175
176    /* pre rotation */
177    for(i=0;i<n8;i++) {
178        re = RSCALE(-input[2*i+n3] - input[n3-1-2*i]);
179        im = RSCALE(-input[n4+2*i] + input[n4-1-2*i]);
180        j = revtab[i];
181        CMUL(x[j].re, x[j].im, re, im, -tcos[i], tsin[i]);
182
183        re = RSCALE( input[2*i]    - input[n2-1-2*i]);
184        im = RSCALE(-input[n2+2*i] - input[ n-1-2*i]);
185        j = revtab[n8 + i];
186        CMUL(x[j].re, x[j].im, re, im, -tcos[n8 + i], tsin[n8 + i]);
187    }
188
189    s->fft_calc(s, x);
190
191    /* post rotation */
192    for(i=0;i<n8;i++) {
193        FFTSample r0, i0, r1, i1;
194        CMUL(i1, r0, x[n8-i-1].re, x[n8-i-1].im, -tsin[n8-i-1], -tcos[n8-i-1]);
195        CMUL(i0, r1, x[n8+i  ].re, x[n8+i  ].im, -tsin[n8+i  ], -tcos[n8+i  ]);
196        x[n8-i-1].re = r0;
197        x[n8-i-1].im = i0;
198        x[n8+i  ].re = r1;
199        x[n8+i  ].im = i1;
200    }
201}
202
203av_cold void ff_mdct_end(FFTContext *s)
204{
205    av_freep(&s->tcos);
206    ff_fft_end(s);
207}
208