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/libm.h"
26#include "libavutil/mathematics.h"
27#include "fft.h"
28#include "fft-internal.h"
29
30/**
31 * @file
32 * MDCT/IMDCT transforms.
33 */
34
35#if FFT_FLOAT
36#   define RSCALE(x, y) ((x) + (y))
37#else
38#   define RSCALE(x, y) ((int)((x) + (unsigned)(y) + 32) >> 6)
39#endif
40
41/**
42 * init MDCT or IMDCT computation.
43 */
44av_cold int ff_mdct_init(FFTContext *s, int nbits, int inverse, double scale)
45{
46    int n, n4, i;
47    double alpha, theta;
48    int tstep;
49
50    memset(s, 0, sizeof(*s));
51    n = 1 << nbits;
52    s->mdct_bits = nbits;
53    s->mdct_size = n;
54    n4 = n >> 2;
55    s->mdct_permutation = FF_MDCT_PERM_NONE;
56
57    if (ff_fft_init(s, s->mdct_bits - 2, inverse) < 0)
58        goto fail;
59
60    s->tcos = av_malloc_array(n/2, sizeof(FFTSample));
61    if (!s->tcos)
62        goto fail;
63
64    switch (s->mdct_permutation) {
65    case FF_MDCT_PERM_NONE:
66        s->tsin = s->tcos + n4;
67        tstep = 1;
68        break;
69    case FF_MDCT_PERM_INTERLEAVE:
70        s->tsin = s->tcos + 1;
71        tstep = 2;
72        break;
73    default:
74        goto fail;
75    }
76
77    theta = 1.0 / 8.0 + (scale < 0 ? n4 : 0);
78    scale = sqrt(fabs(scale));
79    for(i=0;i<n4;i++) {
80        alpha = 2 * M_PI * (i + theta) / n;
81#if !FFT_FLOAT
82        s->tcos[i*tstep] = lrint(-cos(alpha) * 2147483648.0);
83        s->tsin[i*tstep] = lrint(-sin(alpha) * 2147483648.0);
84#else
85        s->tcos[i*tstep] = FIX15(-cos(alpha) * scale);
86        s->tsin[i*tstep] = FIX15(-sin(alpha) * scale);
87#endif
88    }
89    return 0;
90 fail:
91    ff_mdct_end(s);
92    return -1;
93}
94
95/**
96 * Compute the middle half of the inverse MDCT of size N = 2^nbits,
97 * thus excluding the parts that can be derived by symmetry
98 * @param output N/2 samples
99 * @param input N/2 samples
100 */
101void ff_imdct_half_c(FFTContext *s, FFTSample *output, const FFTSample *input)
102{
103    int k, n8, n4, n2, n, j;
104    const uint16_t *revtab = s->revtab;
105    const FFTSample *tcos = s->tcos;
106    const FFTSample *tsin = s->tsin;
107    const FFTSample *in1, *in2;
108    FFTComplex *z = (FFTComplex *)output;
109
110    n = 1 << s->mdct_bits;
111    n2 = n >> 1;
112    n4 = n >> 2;
113    n8 = n >> 3;
114
115    /* pre rotation */
116    in1 = input;
117    in2 = input + n2 - 1;
118    for(k = 0; k < n4; k++) {
119        j=revtab[k];
120        CMUL(z[j].re, z[j].im, *in2, *in1, tcos[k], tsin[k]);
121        in1 += 2;
122        in2 -= 2;
123    }
124    s->fft_calc(s, z);
125
126    /* post rotation + reordering */
127    for(k = 0; k < n8; k++) {
128        FFTSample r0, i0, r1, i1;
129        CMUL(r0, i1, z[n8-k-1].im, z[n8-k-1].re, tsin[n8-k-1], tcos[n8-k-1]);
130        CMUL(r1, i0, z[n8+k  ].im, z[n8+k  ].re, tsin[n8+k  ], tcos[n8+k  ]);
131        z[n8-k-1].re = r0;
132        z[n8-k-1].im = i0;
133        z[n8+k  ].re = r1;
134        z[n8+k  ].im = i1;
135    }
136}
137
138/**
139 * Compute inverse MDCT of size N = 2^nbits
140 * @param output N samples
141 * @param input N/2 samples
142 */
143void ff_imdct_calc_c(FFTContext *s, FFTSample *output, const FFTSample *input)
144{
145    int k;
146    int n = 1 << s->mdct_bits;
147    int n2 = n >> 1;
148    int n4 = n >> 2;
149
150    ff_imdct_half_c(s, output+n4, input);
151
152    for(k = 0; k < n4; k++) {
153        output[k] = -output[n2-k-1];
154        output[n-k-1] = output[n2+k];
155    }
156}
157
158/**
159 * Compute MDCT of size N = 2^nbits
160 * @param input N samples
161 * @param out N/2 samples
162 */
163void ff_mdct_calc_c(FFTContext *s, FFTSample *out, const FFTSample *input)
164{
165    int i, j, n, n8, n4, n2, n3;
166    FFTDouble re, im;
167    const uint16_t *revtab = s->revtab;
168    const FFTSample *tcos = s->tcos;
169    const FFTSample *tsin = s->tsin;
170    FFTComplex *x = (FFTComplex *)out;
171
172    n = 1 << s->mdct_bits;
173    n2 = n >> 1;
174    n4 = n >> 2;
175    n8 = n >> 3;
176    n3 = 3 * n4;
177
178    /* pre rotation */
179    for(i=0;i<n8;i++) {
180        re = RSCALE(-input[2*i+n3], - input[n3-1-2*i]);
181        im = RSCALE(-input[n4+2*i], + input[n4-1-2*i]);
182        j = revtab[i];
183        CMUL(x[j].re, x[j].im, re, im, -tcos[i], tsin[i]);
184
185        re = RSCALE( input[2*i]   , - input[n2-1-2*i]);
186        im = RSCALE(-input[n2+2*i], - input[ n-1-2*i]);
187        j = revtab[n8 + i];
188        CMUL(x[j].re, x[j].im, re, im, -tcos[n8 + i], tsin[n8 + i]);
189    }
190
191    s->fft_calc(s, x);
192
193    /* post rotation */
194    for(i=0;i<n8;i++) {
195        FFTSample r0, i0, r1, i1;
196        CMUL(i1, r0, x[n8-i-1].re, x[n8-i-1].im, -tsin[n8-i-1], -tcos[n8-i-1]);
197        CMUL(i0, r1, x[n8+i  ].re, x[n8+i  ].im, -tsin[n8+i  ], -tcos[n8+i  ]);
198        x[n8-i-1].re = r0;
199        x[n8-i-1].im = i0;
200        x[n8+i  ].re = r1;
201        x[n8+i  ].im = i1;
202    }
203}
204
205av_cold void ff_mdct_end(FFTContext *s)
206{
207    av_freep(&s->tcos);
208    ff_fft_end(s);
209}
210