| @@ -0,0 +1,303 @@ |
| | 1 | /* |
| | 2 | * Bink Audio decoder |
| | 3 | * Copyright (c) 2007-2010 Peter Ross (pross@xvid.org) |
| | 4 | * Copyright (c) 2009 Daniel Verkamp (daniel@drv.nu) |
| | 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 | /** |
| | 24 | * @file libavcodec/binkaudio.c |
| | 25 | * Bink Audio decoder |
| | 26 | * |
| | 27 | * Technical details here: |
| | 28 | * http://wiki.multimedia.cx/index.php?title=Bink_Audio |
| | 29 | */ |
| | 30 | |
| | 31 | #include "avcodec.h" |
| | 32 | #define ALT_BITSTREAM_READER_LE |
| | 33 | #include "get_bits.h" |
| | 34 | #include "dsputil.h" |
| | 35 | extern const uint16_t ff_wma_critical_freqs[25]; |
| | 36 | |
| | 37 | #define MAX_CHANNELS 2 |
| | 38 | #define BINK_BLOCK_MAX_SIZE (MAX_CHANNELS << 11) |
| | 39 | |
| | 40 | typedef struct { |
| | 41 | AVCodecContext *avctx; |
| | 42 | GetBitContext gb; |
| | 43 | DSPContext dsp; |
| | 44 | int first; |
| | 45 | int channels; |
| | 46 | int frame_len; ///< transform size (samples) |
| | 47 | int overlap_len; ///< overlap size (samples) |
| | 48 | int block_size; |
| | 49 | int num_bands; |
| | 50 | unsigned int *bands; |
| | 51 | float root; |
| | 52 | DECLARE_ALIGNED_16(FFTSample, coeffs[BINK_BLOCK_MAX_SIZE]); |
| | 53 | DECLARE_ALIGNED_16(short, previous[BINK_BLOCK_MAX_SIZE / 16]); ///< coeffs from previous audio block |
| | 54 | float *coeffs_ptr[MAX_CHANNELS]; ///< pointers to the coeffs arrays for float_to_int16_interleave |
| | 55 | union { |
| | 56 | RDFTContext rdft; |
| | 57 | DCTContext dct; |
| | 58 | } trans; |
| | 59 | } BinkAudioContext; |
| | 60 | |
| | 61 | |
| | 62 | static av_cold int decode_init(AVCodecContext *avctx) |
| | 63 | { |
| | 64 | BinkAudioContext *s = avctx->priv_data; |
| | 65 | int sample_rate = avctx->sample_rate; |
| | 66 | int sample_rate_half; |
| | 67 | int i; |
| | 68 | int frame_len_bits; |
| | 69 | |
| | 70 | s->avctx = avctx; |
| | 71 | dsputil_init(&s->dsp, avctx); |
| | 72 | |
| | 73 | /* determine frame length */ |
| | 74 | if (avctx->sample_rate < 22050) { |
| | 75 | frame_len_bits = 9; |
| | 76 | } else if (avctx->sample_rate < 44100) { |
| | 77 | frame_len_bits = 10; |
| | 78 | } else { |
| | 79 | frame_len_bits = 11; |
| | 80 | } |
| | 81 | s->frame_len = 1 << frame_len_bits; |
| | 82 | |
| | 83 | if (s->channels > MAX_CHANNELS) { |
| | 84 | av_log(s->avctx, AV_LOG_ERROR, "too many channels: %d\n", s->channels); |
| | 85 | return -1; |
| | 86 | } |
| | 87 | |
| | 88 | if (avctx->codec->id == CODEC_ID_BINKAUDIO_RDFT) { |
| | 89 | // audio is already interleaved for the RDFT format variant |
| | 90 | sample_rate *= avctx->channels; |
| | 91 | s->frame_len *= avctx->channels; |
| | 92 | s->channels = 1; |
| | 93 | if (avctx->channels == 2) |
| | 94 | frame_len_bits++; |
| | 95 | } else { |
| | 96 | s->channels = avctx->channels; |
| | 97 | } |
| | 98 | |
| | 99 | s->overlap_len = s->frame_len / 16; |
| | 100 | s->block_size = (s->frame_len - s->overlap_len) * s->channels; |
| | 101 | sample_rate_half = (sample_rate + 1) / 2; |
| | 102 | s->root = 2.0 / sqrt(s->frame_len); |
| | 103 | |
| | 104 | /* calculate number of bands */ |
| | 105 | for (s->num_bands = 1; s->num_bands < 25; s->num_bands++) |
| | 106 | if (sample_rate_half <= ff_wma_critical_freqs[s->num_bands - 1]) |
| | 107 | break; |
| | 108 | |
| | 109 | s->bands = av_malloc((s->num_bands + 1) * sizeof(*s->bands)); |
| | 110 | if (!s->bands) |
| | 111 | return AVERROR(ENOMEM); |
| | 112 | |
| | 113 | /* populate bands data */ |
| | 114 | s->bands[0] = 1; |
| | 115 | for (i = 1; i < s->num_bands; i++) |
| | 116 | s->bands[i] = ff_wma_critical_freqs[i - 1] * (s->frame_len / 2) / sample_rate_half; |
| | 117 | s->bands[s->num_bands] = s->frame_len / 2; |
| | 118 | |
| | 119 | s->first = 1; |
| | 120 | avctx->sample_fmt = SAMPLE_FMT_S16; |
| | 121 | |
| | 122 | for (i = 0; i < s->channels; i++) |
| | 123 | s->coeffs_ptr[i] = s->coeffs + i * s->frame_len; |
| | 124 | |
| | 125 | if (avctx->codec->id == CODEC_ID_BINKAUDIO_RDFT) |
| | 126 | ff_rdft_init(&s->trans.rdft, frame_len_bits, IRIDFT); |
| | 127 | else |
| | 128 | ff_dct_init(&s->trans.dct, frame_len_bits, 0); |
| | 129 | |
| | 130 | return 0; |
| | 131 | } |
| | 132 | |
| | 133 | static float get_float(GetBitContext *gb) |
| | 134 | { |
| | 135 | int power = get_bits(gb, 5); |
| | 136 | float f = ldexpf(get_bits_long(gb, 23), power - 23); |
| | 137 | if (get_bits1(gb)) |
| | 138 | f = -f; |
| | 139 | return f; |
| | 140 | } |
| | 141 | |
| | 142 | static const uint8_t rle_length_tab[16] = { |
| | 143 | 2, 3, 4, 5, 6, 8, 9, 10, 11, 12, 13, 14, 15, 16, 32, 64 |
| | 144 | }; |
| | 145 | |
| | 146 | /** |
| | 147 | * Decode Bink Audio block |
| | 148 | * @param[out] out Output buffer (must contain s->block_size elements) |
| | 149 | */ |
| | 150 | static void decode_block(BinkAudioContext *s, short *out, int use_dct) |
| | 151 | { |
| | 152 | int ch, i, j, k; |
| | 153 | float q, quant[25]; |
| | 154 | int width, coeff; |
| | 155 | GetBitContext *gb = &s->gb; |
| | 156 | |
| | 157 | if (use_dct) |
| | 158 | skip_bits(gb, 2); |
| | 159 | |
| | 160 | for (ch = 0; ch < s->channels; ch++) { |
| | 161 | FFTSample *coeffs = s->coeffs_ptr[ch]; |
| | 162 | q = 0.0f; |
| | 163 | coeffs[0] = get_float(gb) * s->root; |
| | 164 | coeffs[1] = get_float(gb) * s->root; |
| | 165 | |
| | 166 | for (i = 0; i < s->num_bands; i++) { |
| | 167 | /* constant is result of 0.066399999/log10(M_E) */ |
| | 168 | int value = get_bits(gb, 8); |
| | 169 | quant[i] = expf(FFMIN(value, 95) * 0.15289164787221953823f) * s->root; |
| | 170 | } |
| | 171 | |
| | 172 | // find band (k) |
| | 173 | for (k = 0; s->bands[k] < 1; k++) { |
| | 174 | q = quant[k]; |
| | 175 | } |
| | 176 | |
| | 177 | // parse coefficients |
| | 178 | i = 2; |
| | 179 | while (i < s->frame_len) { |
| | 180 | if (get_bits1(gb)) { |
| | 181 | j = i + rle_length_tab[get_bits(gb, 4)] * 8; |
| | 182 | } else { |
| | 183 | j = i + 8; |
| | 184 | } |
| | 185 | |
| | 186 | j = FFMIN(j, s->frame_len); |
| | 187 | |
| | 188 | width = get_bits(gb, 4); |
| | 189 | if (width == 0) { |
| | 190 | memset(coeffs + i, 0, (j - i) * sizeof(*coeffs)); |
| | 191 | i = j; |
| | 192 | while (s->bands[k] * 2 < i) |
| | 193 | q = quant[k++]; |
| | 194 | } else { |
| | 195 | while (i < j) { |
| | 196 | if (s->bands[k] * 2 == i) |
| | 197 | q = quant[k++]; |
| | 198 | coeff = get_bits(gb, width); |
| | 199 | if (coeff) { |
| | 200 | if (get_bits1(gb)) |
| | 201 | coeffs[i] = -q * coeff; |
| | 202 | else |
| | 203 | coeffs[i] = q * coeff; |
| | 204 | } else { |
| | 205 | coeffs[i] = 0.0f; |
| | 206 | } |
| | 207 | i++; |
| | 208 | } |
| | 209 | } |
| | 210 | } |
| | 211 | |
| | 212 | if (use_dct) |
| | 213 | ff_dct_calc (&s->trans.dct, coeffs); |
| | 214 | else |
| | 215 | ff_rdft_calc(&s->trans.rdft, coeffs); |
| | 216 | } |
| | 217 | |
| | 218 | s->dsp.float_to_int16_interleave(out, (const float **)s->coeffs_ptr, s->frame_len, s->channels); |
| | 219 | |
| | 220 | if (!s->first) { |
| | 221 | int count = s->overlap_len * s->channels; |
| | 222 | int shift = av_log2(count); |
| | 223 | for (i = 0; i < count; i++) { |
| | 224 | out[i] = (s->previous[i] * (count - i) + out[i] * i) >> shift; |
| | 225 | } |
| | 226 | } |
| | 227 | |
| | 228 | memcpy(s->previous, out + s->block_size, |
| | 229 | s->overlap_len * s->channels * sizeof(*out)); |
| | 230 | |
| | 231 | s->first = 0; |
| | 232 | } |
| | 233 | |
| | 234 | static av_cold int decode_end(AVCodecContext *avctx) |
| | 235 | { |
| | 236 | BinkAudioContext * s = avctx->priv_data; |
| | 237 | av_freep(&s->bands); |
| | 238 | if (avctx->codec->id == CODEC_ID_BINKAUDIO_RDFT) |
| | 239 | ff_rdft_end(&s->trans.rdft); |
| | 240 | else |
| | 241 | ff_dct_end(&s->trans.dct); |
| | 242 | return 0; |
| | 243 | } |
| | 244 | |
| | 245 | static void get_bits_align32(GetBitContext *s) |
| | 246 | { |
| | 247 | int n = (-get_bits_count(s)) & 31; |
| | 248 | if (n) skip_bits(s, n); |
| | 249 | } |
| | 250 | |
| | 251 | static int decode_frame(AVCodecContext *avctx, |
| | 252 | void *data, int *data_size, |
| | 253 | AVPacket *avpkt) |
| | 254 | { |
| | 255 | BinkAudioContext *s = avctx->priv_data; |
| | 256 | const uint8_t *buf = avpkt->data; |
| | 257 | int buf_size = avpkt->size; |
| | 258 | short *samples = data; |
| | 259 | short *samples_end = (short*)((uint8_t*)data + *data_size); |
| | 260 | int reported_size; |
| | 261 | GetBitContext *gb = &s->gb; |
| | 262 | |
| | 263 | init_get_bits(gb, buf, buf_size * 8); |
| | 264 | |
| | 265 | reported_size = get_bits_long(gb, 32); |
| | 266 | while (get_bits_count(gb) / 8 < buf_size && |
| | 267 | samples + s->block_size <= samples_end) { |
| | 268 | decode_block(s, samples, avctx->codec->id == CODEC_ID_BINKAUDIO_DCT); |
| | 269 | samples += s->block_size; |
| | 270 | get_bits_align32(gb); |
| | 271 | } |
| | 272 | |
| | 273 | *data_size = (uint8_t*)samples - (uint8_t*)data; |
| | 274 | if (reported_size != *data_size) { |
| | 275 | av_log(avctx, AV_LOG_WARNING, "reported data size (%d) does not match output data size (%d)\n", |
| | 276 | reported_size, *data_size); |
| | 277 | } |
| | 278 | return buf_size; |
| | 279 | } |
| | 280 | |
| | 281 | AVCodec binkaudio_rdft_decoder = { |
| | 282 | "binkaudio_rdft", |
| | 283 | CODEC_TYPE_AUDIO, |
| | 284 | CODEC_ID_BINKAUDIO_RDFT, |
| | 285 | sizeof(BinkAudioContext), |
| | 286 | decode_init, |
| | 287 | NULL, |
| | 288 | decode_end, |
| | 289 | decode_frame, |
| | 290 | .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (RDFT)") |
| | 291 | }; |
| | 292 | |
| | 293 | AVCodec binkaudio_dct_decoder = { |
| | 294 | "binkaudio_dct", |
| | 295 | CODEC_TYPE_AUDIO, |
| | 296 | CODEC_ID_BINKAUDIO_DCT, |
| | 297 | sizeof(BinkAudioContext), |
| | 298 | decode_init, |
| | 299 | NULL, |
| | 300 | decode_end, |
| | 301 | decode_frame, |
| | 302 | .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (DCT)") |
| | 303 | }; |