X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fmdct.c;h=6f645342732a018a5157fa4e0ac2a02df739770d;hb=cac4760bf42f1b7ec93860399fbb7b5bff90005f;hp=9de5c2415996665b7371eacb0fe8304a88f0a52c;hpb=bad5537e2c2caeb5deb1ff9d771ea01058b8010c;p=ffmpeg diff --git a/libavcodec/mdct.c b/libavcodec/mdct.c index 9de5c241599..6f645342732 100644 --- a/libavcodec/mdct.c +++ b/libavcodec/mdct.c @@ -2,130 +2,106 @@ * MDCT/IMDCT transforms * Copyright (c) 2002 Fabrice Bellard * - * This file is part of FFmpeg. + * This file is part of Libav. * - * FFmpeg is free software; you can redistribute it and/or + * Libav is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * - * FFmpeg is distributed in the hope that it will be useful, + * Libav is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public - * License along with FFmpeg; if not, write to the Free Software + * License along with Libav; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ -#include "dsputil.h" + +#include +#include +#include "libavutil/common.h" +#include "libavutil/mathematics.h" +#include "fft.h" +#include "fft-internal.h" /** - * @file libavcodec/mdct.c + * @file * MDCT/IMDCT transforms. */ -// Generate a Kaiser-Bessel Derived Window. -#define BESSEL_I0_ITER 50 // default: 50 iterations of Bessel I0 approximation -void ff_kbd_window_init(float *window, float alpha, int n) -{ - int i, j; - double sum = 0.0, bessel, tmp; - double local_window[n]; - double alpha2 = (alpha * M_PI / n) * (alpha * M_PI / n); - - for (i = 0; i < n; i++) { - tmp = i * (n - i) * alpha2; - bessel = 1.0; - for (j = BESSEL_I0_ITER; j > 0; j--) - bessel = bessel * tmp / (j * j) + 1; - sum += bessel; - local_window[i] = sum; - } - - sum++; - for (i = 0; i < n; i++) - window[i] = sqrt(local_window[i] / sum); -} - -DECLARE_ALIGNED(16, float, ff_sine_128 [ 128]); -DECLARE_ALIGNED(16, float, ff_sine_256 [ 256]); -DECLARE_ALIGNED(16, float, ff_sine_512 [ 512]); -DECLARE_ALIGNED(16, float, ff_sine_1024[1024]); -DECLARE_ALIGNED(16, float, ff_sine_2048[2048]); -DECLARE_ALIGNED(16, float, ff_sine_4096[4096]); -float *ff_sine_windows[6] = { - ff_sine_128, ff_sine_256, ff_sine_512, ff_sine_1024, ff_sine_2048, ff_sine_4096 -}; - -// Generate a sine window. -void ff_sine_window_init(float *window, int n) { - int i; - for(i = 0; i < n; i++) - window[i] = sinf((i + 0.5) * (M_PI / (2.0 * n))); -} +#if CONFIG_FFT_FLOAT +# define RSCALE(x) (x) +#else +# define RSCALE(x) ((x) >> 1) +#endif /** * init MDCT or IMDCT computation. */ -int ff_mdct_init(MDCTContext *s, int nbits, int inverse) +av_cold int ff_mdct_init(FFTContext *s, int nbits, int inverse, double scale) { int n, n4, i; - double alpha; + double alpha, theta; + int tstep; memset(s, 0, sizeof(*s)); n = 1 << nbits; - s->nbits = nbits; - s->n = n; + s->mdct_bits = nbits; + s->mdct_size = n; n4 = n >> 2; - s->tcos = av_malloc(n4 * sizeof(FFTSample)); + s->mdct_permutation = FF_MDCT_PERM_NONE; + + if (ff_fft_init(s, s->mdct_bits - 2, inverse) < 0) + goto fail; + + s->tcos = av_malloc(n/2 * sizeof(FFTSample)); if (!s->tcos) goto fail; - s->tsin = av_malloc(n4 * sizeof(FFTSample)); - if (!s->tsin) + + switch (s->mdct_permutation) { + case FF_MDCT_PERM_NONE: + s->tsin = s->tcos + n4; + tstep = 1; + break; + case FF_MDCT_PERM_INTERLEAVE: + s->tsin = s->tcos + 1; + tstep = 2; + break; + default: goto fail; + } + theta = 1.0 / 8.0 + (scale < 0 ? n4 : 0); + scale = sqrt(fabs(scale)); for(i=0;itcos[i] = -cos(alpha); - s->tsin[i] = -sin(alpha); + alpha = 2 * M_PI * (i + theta) / n; + s->tcos[i*tstep] = FIX15(-cos(alpha) * scale); + s->tsin[i*tstep] = FIX15(-sin(alpha) * scale); } - if (ff_fft_init(&s->fft, s->nbits - 2, inverse) < 0) - goto fail; return 0; fail: - av_freep(&s->tcos); - av_freep(&s->tsin); + ff_mdct_end(s); return -1; } -/* complex multiplication: p = a * b */ -#define CMUL(pre, pim, are, aim, bre, bim) \ -{\ - FFTSample _are = (are);\ - FFTSample _aim = (aim);\ - FFTSample _bre = (bre);\ - FFTSample _bim = (bim);\ - (pre) = _are * _bre - _aim * _bim;\ - (pim) = _are * _bim + _aim * _bre;\ -} - /** * Compute the middle half of the inverse MDCT of size N = 2^nbits, * thus excluding the parts that can be derived by symmetry * @param output N/2 samples * @param input N/2 samples */ -void ff_imdct_half_c(MDCTContext *s, FFTSample *output, const FFTSample *input) +void ff_imdct_half_c(FFTContext *s, FFTSample *output, const FFTSample *input) { int k, n8, n4, n2, n, j; - const uint16_t *revtab = s->fft.revtab; + const uint16_t *revtab = s->revtab; const FFTSample *tcos = s->tcos; const FFTSample *tsin = s->tsin; const FFTSample *in1, *in2; FFTComplex *z = (FFTComplex *)output; - n = 1 << s->nbits; + n = 1 << s->mdct_bits; n2 = n >> 1; n4 = n >> 2; n8 = n >> 3; @@ -139,10 +115,9 @@ void ff_imdct_half_c(MDCTContext *s, FFTSample *output, const FFTSample *input) in1 += 2; in2 -= 2; } - ff_fft_calc(&s->fft, z); + s->fft_calc(s, z); /* post rotation + reordering */ - output += n4; for(k = 0; k < n8; k++) { FFTSample r0, i0, r1, i1; CMUL(r0, i1, z[n8-k-1].im, z[n8-k-1].re, tsin[n8-k-1], tcos[n8-k-1]); @@ -158,12 +133,11 @@ void ff_imdct_half_c(MDCTContext *s, FFTSample *output, const FFTSample *input) * Compute inverse MDCT of size N = 2^nbits * @param output N samples * @param input N/2 samples - * @param tmp N/2 samples */ -void ff_imdct_calc_c(MDCTContext *s, FFTSample *output, const FFTSample *input) +void ff_imdct_calc_c(FFTContext *s, FFTSample *output, const FFTSample *input) { int k; - int n = 1 << s->nbits; + int n = 1 << s->mdct_bits; int n2 = n >> 1; int n4 = n >> 2; @@ -179,18 +153,17 @@ void ff_imdct_calc_c(MDCTContext *s, FFTSample *output, const FFTSample *input) * Compute MDCT of size N = 2^nbits * @param input N samples * @param out N/2 samples - * @param tmp temporary storage of N/2 samples */ -void ff_mdct_calc(MDCTContext *s, FFTSample *out, const FFTSample *input) +void ff_mdct_calc_c(FFTContext *s, FFTSample *out, const FFTSample *input) { int i, j, n, n8, n4, n2, n3; - FFTSample re, im; - const uint16_t *revtab = s->fft.revtab; + FFTDouble re, im; + const uint16_t *revtab = s->revtab; const FFTSample *tcos = s->tcos; const FFTSample *tsin = s->tsin; FFTComplex *x = (FFTComplex *)out; - n = 1 << s->nbits; + n = 1 << s->mdct_bits; n2 = n >> 1; n4 = n >> 2; n8 = n >> 3; @@ -198,18 +171,18 @@ void ff_mdct_calc(MDCTContext *s, FFTSample *out, const FFTSample *input) /* pre rotation */ for(i=0;ifft, x); + s->fft_calc(s, x); /* post rotation */ for(i=0;itcos); - av_freep(&s->tsin); - ff_fft_end(&s->fft); + ff_fft_end(s); }