ac3dsp.c 7.68 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
/*
 * AC-3 DSP utils
 * Copyright (c) 2011 Justin Ruggles
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg 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,
 * 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
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

22
#include "libavutil/avassert.h"
23
#include "avcodec.h"
24
#include "ac3.h"
25
#include "ac3dsp.h"
26
#include "mathops.h"
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

static void ac3_exponent_min_c(uint8_t *exp, int num_reuse_blocks, int nb_coefs)
{
    int blk, i;

    if (!num_reuse_blocks)
        return;

    for (i = 0; i < nb_coefs; i++) {
        uint8_t min_exp = *exp;
        uint8_t *exp1 = exp + 256;
        for (blk = 0; blk < num_reuse_blocks; blk++) {
            uint8_t next_exp = *exp1;
            if (next_exp < min_exp)
                min_exp = next_exp;
            exp1 += 256;
        }
        *exp++ = min_exp;
    }
}

48 49 50 51 52 53 54 55
static int ac3_max_msb_abs_int16_c(const int16_t *src, int len)
{
    int i, v = 0;
    for (i = 0; i < len; i++)
        v |= abs(src[i]);
    return v;
}

56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90
static void ac3_lshift_int16_c(int16_t *src, unsigned int len,
                               unsigned int shift)
{
    uint32_t *src32 = (uint32_t *)src;
    const uint32_t mask = ~(((1 << shift) - 1) << 16);
    int i;
    len >>= 1;
    for (i = 0; i < len; i += 8) {
        src32[i  ] = (src32[i  ] << shift) & mask;
        src32[i+1] = (src32[i+1] << shift) & mask;
        src32[i+2] = (src32[i+2] << shift) & mask;
        src32[i+3] = (src32[i+3] << shift) & mask;
        src32[i+4] = (src32[i+4] << shift) & mask;
        src32[i+5] = (src32[i+5] << shift) & mask;
        src32[i+6] = (src32[i+6] << shift) & mask;
        src32[i+7] = (src32[i+7] << shift) & mask;
    }
}

static void ac3_rshift_int32_c(int32_t *src, unsigned int len,
                               unsigned int shift)
{
    do {
        *src++ >>= shift;
        *src++ >>= shift;
        *src++ >>= shift;
        *src++ >>= shift;
        *src++ >>= shift;
        *src++ >>= shift;
        *src++ >>= shift;
        *src++ >>= shift;
        len -= 8;
    } while (len > 0);
}

91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106
static void float_to_fixed24_c(int32_t *dst, const float *src, unsigned int len)
{
    const float scale = 1 << 24;
    do {
        *dst++ = lrintf(*src++ * scale);
        *dst++ = lrintf(*src++ * scale);
        *dst++ = lrintf(*src++ * scale);
        *dst++ = lrintf(*src++ * scale);
        *dst++ = lrintf(*src++ * scale);
        *dst++ = lrintf(*src++ * scale);
        *dst++ = lrintf(*src++ * scale);
        *dst++ = lrintf(*src++ * scale);
        len -= 8;
    } while (len > 0);
}

107 108 109 110 111
static void ac3_bit_alloc_calc_bap_c(int16_t *mask, int16_t *psd,
                                     int start, int end,
                                     int snr_offset, int floor,
                                     const uint8_t *bap_tab, uint8_t *bap)
{
112
    int bin, band, band_end;
113 114 115 116 117 118 119 120 121 122 123

    /* special case, if snr offset is -960, set all bap's to zero */
    if (snr_offset == -960) {
        memset(bap, 0, AC3_MAX_COEFS);
        return;
    }

    bin  = start;
    band = ff_ac3_bin_to_band_tab[start];
    do {
        int m = (FFMAX(mask[band] - snr_offset - floor, 0) & 0x1FE0) + floor;
124 125 126
        band_end = ff_ac3_band_start_tab[++band];
        band_end = FFMIN(band_end, end);

127 128 129 130
        for (; bin < band_end; bin++) {
            int address = av_clip((psd[bin] - m) >> 5, 0, 63);
            bap[bin] = bap_tab[address];
        }
131
    } while (end > band_end);
132 133
}

134 135
static void ac3_update_bap_counts_c(uint16_t mant_cnt[16], uint8_t *bap,
                                    int len)
136
{
137
    while (len-- > 0)
138 139
        mant_cnt[bap[len]]++;
}
140

141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
DECLARE_ALIGNED(16, const uint16_t, ff_ac3_bap_bits)[16] = {
    0,  0,  0,  3,  0,  4,  5,  6,  7,  8,  9, 10, 11, 12, 14, 16
};

static int ac3_compute_mantissa_size_c(uint16_t mant_cnt[6][16])
{
    int blk, bap;
    int bits = 0;

    for (blk = 0; blk < AC3_MAX_BLOCKS; blk++) {
        // bap=1 : 3 mantissas in 5 bits
        bits += (mant_cnt[blk][1] / 3) * 5;
        // bap=2 : 3 mantissas in 7 bits
        // bap=4 : 2 mantissas in 7 bits
        bits += ((mant_cnt[blk][2] / 3) + (mant_cnt[blk][4] >> 1)) * 7;
        // bap=3 : 1 mantissa in 3 bits
        bits += mant_cnt[blk][3] * 3;
        // bap=5 to 15 : get bits per mantissa from table
        for (bap = 5; bap < 16; bap++)
            bits += mant_cnt[blk][bap] * ff_ac3_bap_bits[bap];
161 162 163 164
    }
    return bits;
}

165 166 167 168 169 170
static void ac3_extract_exponents_c(uint8_t *exp, int32_t *coef, int nb_coefs)
{
    int i;

    for (i = 0; i < nb_coefs; i++) {
        int v = abs(coef[i]);
171
        exp[i] = v ? 23 - av_log2(v) : 24;
172 173 174
    }
}

175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216
static void ac3_sum_square_butterfly_int32_c(int64_t sum[4],
                                             const int32_t *coef0,
                                             const int32_t *coef1,
                                             int len)
{
    int i;

    sum[0] = sum[1] = sum[2] = sum[3] = 0;

    for (i = 0; i < len; i++) {
        int lt = coef0[i];
        int rt = coef1[i];
        int md = lt + rt;
        int sd = lt - rt;
        MAC64(sum[0], lt, lt);
        MAC64(sum[1], rt, rt);
        MAC64(sum[2], md, md);
        MAC64(sum[3], sd, sd);
    }
}

static void ac3_sum_square_butterfly_float_c(float sum[4],
                                             const float *coef0,
                                             const float *coef1,
                                             int len)
{
    int i;

    sum[0] = sum[1] = sum[2] = sum[3] = 0;

    for (i = 0; i < len; i++) {
        float lt = coef0[i];
        float rt = coef1[i];
        float md = lt + rt;
        float sd = lt - rt;
        sum[0] += lt * lt;
        sum[1] += rt * rt;
        sum[2] += md * md;
        sum[3] += sd * sd;
    }
}

217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241
static void ac3_downmix_c(float (*samples)[256], float (*matrix)[2],
                          int out_ch, int in_ch, int len)
{
    int i, j;
    float v0, v1;
    if (out_ch == 2) {
        for (i = 0; i < len; i++) {
            v0 = v1 = 0.0f;
            for (j = 0; j < in_ch; j++) {
                v0 += samples[j][i] * matrix[j][0];
                v1 += samples[j][i] * matrix[j][1];
            }
            samples[0][i] = v0;
            samples[1][i] = v1;
        }
    } else if (out_ch == 1) {
        for (i = 0; i < len; i++) {
            v0 = 0.0f;
            for (j = 0; j < in_ch; j++)
                v0 += samples[j][i] * matrix[j][0];
            samples[0][i] = v0;
        }
    }
}

242
av_cold void ff_ac3dsp_init(AC3DSPContext *c, int bit_exact)
243 244
{
    c->ac3_exponent_min = ac3_exponent_min_c;
245
    c->ac3_max_msb_abs_int16 = ac3_max_msb_abs_int16_c;
246 247
    c->ac3_lshift_int16 = ac3_lshift_int16_c;
    c->ac3_rshift_int32 = ac3_rshift_int32_c;
248
    c->float_to_fixed24 = float_to_fixed24_c;
249
    c->bit_alloc_calc_bap = ac3_bit_alloc_calc_bap_c;
250
    c->update_bap_counts = ac3_update_bap_counts_c;
251
    c->compute_mantissa_size = ac3_compute_mantissa_size_c;
252
    c->extract_exponents = ac3_extract_exponents_c;
253 254
    c->sum_square_butterfly_int32 = ac3_sum_square_butterfly_int32_c;
    c->sum_square_butterfly_float = ac3_sum_square_butterfly_float_c;
255
    c->downmix = ac3_downmix_c;
256

257 258
    if (ARCH_ARM)
        ff_ac3dsp_init_arm(c, bit_exact);
259
    if (ARCH_X86)
260
        ff_ac3dsp_init_x86(c, bit_exact);
261
}