Go to most recent revision | Details | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
4349 | Serge | 1 | /* |
2 | * Bink Audio decoder |
||
3 | * Copyright (c) 2007-2011 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 |
||
25 | * Bink Audio decoder |
||
26 | * |
||
27 | * Technical details here: |
||
28 | * http://wiki.multimedia.cx/index.php?title=Bink_Audio |
||
29 | */ |
||
30 | |||
31 | #include "libavutil/channel_layout.h" |
||
32 | #include "avcodec.h" |
||
33 | #define BITSTREAM_READER_LE |
||
34 | #include "get_bits.h" |
||
35 | #include "dct.h" |
||
36 | #include "rdft.h" |
||
37 | #include "fmtconvert.h" |
||
38 | #include "internal.h" |
||
39 | #include "wma.h" |
||
40 | #include "libavutil/intfloat.h" |
||
41 | |||
42 | static float quant_table[96]; |
||
43 | |||
44 | #define MAX_CHANNELS 2 |
||
45 | #define BINK_BLOCK_MAX_SIZE (MAX_CHANNELS << 11) |
||
46 | |||
47 | typedef struct { |
||
48 | GetBitContext gb; |
||
49 | int version_b; ///< Bink version 'b' |
||
50 | int first; |
||
51 | int channels; |
||
52 | int frame_len; ///< transform size (samples) |
||
53 | int overlap_len; ///< overlap size (samples) |
||
54 | int block_size; |
||
55 | int num_bands; |
||
56 | unsigned int *bands; |
||
57 | float root; |
||
58 | DECLARE_ALIGNED(32, FFTSample, coeffs)[BINK_BLOCK_MAX_SIZE]; |
||
59 | float previous[MAX_CHANNELS][BINK_BLOCK_MAX_SIZE / 16]; ///< coeffs from previous audio block |
||
60 | uint8_t *packet_buffer; |
||
61 | union { |
||
62 | RDFTContext rdft; |
||
63 | DCTContext dct; |
||
64 | } trans; |
||
65 | } BinkAudioContext; |
||
66 | |||
67 | |||
68 | static av_cold int decode_init(AVCodecContext *avctx) |
||
69 | { |
||
70 | BinkAudioContext *s = avctx->priv_data; |
||
71 | int sample_rate = avctx->sample_rate; |
||
72 | int sample_rate_half; |
||
73 | int i; |
||
74 | int frame_len_bits; |
||
75 | |||
76 | /* determine frame length */ |
||
77 | if (avctx->sample_rate < 22050) { |
||
78 | frame_len_bits = 9; |
||
79 | } else if (avctx->sample_rate < 44100) { |
||
80 | frame_len_bits = 10; |
||
81 | } else { |
||
82 | frame_len_bits = 11; |
||
83 | } |
||
84 | |||
85 | if (avctx->channels < 1 || avctx->channels > MAX_CHANNELS) { |
||
86 | av_log(avctx, AV_LOG_ERROR, "invalid number of channels: %d\n", avctx->channels); |
||
87 | return AVERROR_INVALIDDATA; |
||
88 | } |
||
89 | avctx->channel_layout = avctx->channels == 1 ? AV_CH_LAYOUT_MONO : |
||
90 | AV_CH_LAYOUT_STEREO; |
||
91 | |||
92 | s->version_b = avctx->extradata_size >= 4 && avctx->extradata[3] == 'b'; |
||
93 | |||
94 | if (avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT) { |
||
95 | // audio is already interleaved for the RDFT format variant |
||
96 | avctx->sample_fmt = AV_SAMPLE_FMT_FLT; |
||
97 | sample_rate *= avctx->channels; |
||
98 | s->channels = 1; |
||
99 | if (!s->version_b) |
||
100 | frame_len_bits += av_log2(avctx->channels); |
||
101 | } else { |
||
102 | s->channels = avctx->channels; |
||
103 | avctx->sample_fmt = AV_SAMPLE_FMT_FLTP; |
||
104 | } |
||
105 | |||
106 | s->frame_len = 1 << frame_len_bits; |
||
107 | s->overlap_len = s->frame_len / 16; |
||
108 | s->block_size = (s->frame_len - s->overlap_len) * s->channels; |
||
109 | sample_rate_half = (sample_rate + 1) / 2; |
||
110 | if (avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT) |
||
111 | s->root = 2.0 / (sqrt(s->frame_len) * 32768.0); |
||
112 | else |
||
113 | s->root = s->frame_len / (sqrt(s->frame_len) * 32768.0); |
||
114 | for (i = 0; i < 96; i++) { |
||
115 | /* constant is result of 0.066399999/log10(M_E) */ |
||
116 | quant_table[i] = expf(i * 0.15289164787221953823f) * s->root; |
||
117 | } |
||
118 | |||
119 | /* calculate number of bands */ |
||
120 | for (s->num_bands = 1; s->num_bands < 25; s->num_bands++) |
||
121 | if (sample_rate_half <= ff_wma_critical_freqs[s->num_bands - 1]) |
||
122 | break; |
||
123 | |||
124 | s->bands = av_malloc((s->num_bands + 1) * sizeof(*s->bands)); |
||
125 | if (!s->bands) |
||
126 | return AVERROR(ENOMEM); |
||
127 | |||
128 | /* populate bands data */ |
||
129 | s->bands[0] = 2; |
||
130 | for (i = 1; i < s->num_bands; i++) |
||
131 | s->bands[i] = (ff_wma_critical_freqs[i - 1] * s->frame_len / sample_rate_half) & ~1; |
||
132 | s->bands[s->num_bands] = s->frame_len; |
||
133 | |||
134 | s->first = 1; |
||
135 | |||
136 | if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT) |
||
137 | ff_rdft_init(&s->trans.rdft, frame_len_bits, DFT_C2R); |
||
138 | else if (CONFIG_BINKAUDIO_DCT_DECODER) |
||
139 | ff_dct_init(&s->trans.dct, frame_len_bits, DCT_III); |
||
140 | else |
||
141 | return -1; |
||
142 | |||
143 | return 0; |
||
144 | } |
||
145 | |||
146 | static float get_float(GetBitContext *gb) |
||
147 | { |
||
148 | int power = get_bits(gb, 5); |
||
149 | float f = ldexpf(get_bits_long(gb, 23), power - 23); |
||
150 | if (get_bits1(gb)) |
||
151 | f = -f; |
||
152 | return f; |
||
153 | } |
||
154 | |||
155 | static const uint8_t rle_length_tab[16] = { |
||
156 | 2, 3, 4, 5, 6, 8, 9, 10, 11, 12, 13, 14, 15, 16, 32, 64 |
||
157 | }; |
||
158 | |||
159 | /** |
||
160 | * Decode Bink Audio block |
||
161 | * @param[out] out Output buffer (must contain s->block_size elements) |
||
162 | * @return 0 on success, negative error code on failure |
||
163 | */ |
||
164 | static int decode_block(BinkAudioContext *s, float **out, int use_dct) |
||
165 | { |
||
166 | int ch, i, j, k; |
||
167 | float q, quant[25]; |
||
168 | int width, coeff; |
||
169 | GetBitContext *gb = &s->gb; |
||
170 | |||
171 | if (use_dct) |
||
172 | skip_bits(gb, 2); |
||
173 | |||
174 | for (ch = 0; ch < s->channels; ch++) { |
||
175 | FFTSample *coeffs = out[ch]; |
||
176 | |||
177 | if (s->version_b) { |
||
178 | if (get_bits_left(gb) < 64) |
||
179 | return AVERROR_INVALIDDATA; |
||
180 | coeffs[0] = av_int2float(get_bits_long(gb, 32)) * s->root; |
||
181 | coeffs[1] = av_int2float(get_bits_long(gb, 32)) * s->root; |
||
182 | } else { |
||
183 | if (get_bits_left(gb) < 58) |
||
184 | return AVERROR_INVALIDDATA; |
||
185 | coeffs[0] = get_float(gb) * s->root; |
||
186 | coeffs[1] = get_float(gb) * s->root; |
||
187 | } |
||
188 | |||
189 | if (get_bits_left(gb) < s->num_bands * 8) |
||
190 | return AVERROR_INVALIDDATA; |
||
191 | for (i = 0; i < s->num_bands; i++) { |
||
192 | int value = get_bits(gb, 8); |
||
193 | quant[i] = quant_table[FFMIN(value, 95)]; |
||
194 | } |
||
195 | |||
196 | k = 0; |
||
197 | q = quant[0]; |
||
198 | |||
199 | // parse coefficients |
||
200 | i = 2; |
||
201 | while (i < s->frame_len) { |
||
202 | if (s->version_b) { |
||
203 | j = i + 16; |
||
204 | } else { |
||
205 | int v = get_bits1(gb); |
||
206 | if (v) { |
||
207 | v = get_bits(gb, 4); |
||
208 | j = i + rle_length_tab[v] * 8; |
||
209 | } else { |
||
210 | j = i + 8; |
||
211 | } |
||
212 | } |
||
213 | |||
214 | j = FFMIN(j, s->frame_len); |
||
215 | |||
216 | width = get_bits(gb, 4); |
||
217 | if (width == 0) { |
||
218 | memset(coeffs + i, 0, (j - i) * sizeof(*coeffs)); |
||
219 | i = j; |
||
220 | while (s->bands[k] < i) |
||
221 | q = quant[k++]; |
||
222 | } else { |
||
223 | while (i < j) { |
||
224 | if (s->bands[k] == i) |
||
225 | q = quant[k++]; |
||
226 | coeff = get_bits(gb, width); |
||
227 | if (coeff) { |
||
228 | int v; |
||
229 | v = get_bits1(gb); |
||
230 | if (v) |
||
231 | coeffs[i] = -q * coeff; |
||
232 | else |
||
233 | coeffs[i] = q * coeff; |
||
234 | } else { |
||
235 | coeffs[i] = 0.0f; |
||
236 | } |
||
237 | i++; |
||
238 | } |
||
239 | } |
||
240 | } |
||
241 | |||
242 | if (CONFIG_BINKAUDIO_DCT_DECODER && use_dct) { |
||
243 | coeffs[0] /= 0.5; |
||
244 | s->trans.dct.dct_calc(&s->trans.dct, coeffs); |
||
245 | } |
||
246 | else if (CONFIG_BINKAUDIO_RDFT_DECODER) |
||
247 | s->trans.rdft.rdft_calc(&s->trans.rdft, coeffs); |
||
248 | } |
||
249 | |||
250 | for (ch = 0; ch < s->channels; ch++) { |
||
251 | int j; |
||
252 | int count = s->overlap_len * s->channels; |
||
253 | if (!s->first) { |
||
254 | j = ch; |
||
255 | for (i = 0; i < s->overlap_len; i++, j += s->channels) |
||
256 | out[ch][i] = (s->previous[ch][i] * (count - j) + |
||
257 | out[ch][i] * j) / count; |
||
258 | } |
||
259 | memcpy(s->previous[ch], &out[ch][s->frame_len - s->overlap_len], |
||
260 | s->overlap_len * sizeof(*s->previous[ch])); |
||
261 | } |
||
262 | |||
263 | s->first = 0; |
||
264 | |||
265 | return 0; |
||
266 | } |
||
267 | |||
268 | static av_cold int decode_end(AVCodecContext *avctx) |
||
269 | { |
||
270 | BinkAudioContext * s = avctx->priv_data; |
||
271 | av_freep(&s->bands); |
||
272 | av_freep(&s->packet_buffer); |
||
273 | if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT) |
||
274 | ff_rdft_end(&s->trans.rdft); |
||
275 | else if (CONFIG_BINKAUDIO_DCT_DECODER) |
||
276 | ff_dct_end(&s->trans.dct); |
||
277 | |||
278 | return 0; |
||
279 | } |
||
280 | |||
281 | static void get_bits_align32(GetBitContext *s) |
||
282 | { |
||
283 | int n = (-get_bits_count(s)) & 31; |
||
284 | if (n) skip_bits(s, n); |
||
285 | } |
||
286 | |||
287 | static int decode_frame(AVCodecContext *avctx, void *data, |
||
288 | int *got_frame_ptr, AVPacket *avpkt) |
||
289 | { |
||
290 | BinkAudioContext *s = avctx->priv_data; |
||
291 | AVFrame *frame = data; |
||
292 | GetBitContext *gb = &s->gb; |
||
293 | int ret, consumed = 0; |
||
294 | |||
295 | if (!get_bits_left(gb)) { |
||
296 | uint8_t *buf; |
||
297 | /* handle end-of-stream */ |
||
298 | if (!avpkt->size) { |
||
299 | *got_frame_ptr = 0; |
||
300 | return 0; |
||
301 | } |
||
302 | if (avpkt->size < 4) { |
||
303 | av_log(avctx, AV_LOG_ERROR, "Packet is too small\n"); |
||
304 | return AVERROR_INVALIDDATA; |
||
305 | } |
||
306 | buf = av_realloc(s->packet_buffer, avpkt->size + FF_INPUT_BUFFER_PADDING_SIZE); |
||
307 | if (!buf) |
||
308 | return AVERROR(ENOMEM); |
||
309 | s->packet_buffer = buf; |
||
310 | memcpy(s->packet_buffer, avpkt->data, avpkt->size); |
||
311 | init_get_bits(gb, s->packet_buffer, avpkt->size * 8); |
||
312 | consumed = avpkt->size; |
||
313 | |||
314 | /* skip reported size */ |
||
315 | skip_bits_long(gb, 32); |
||
316 | } |
||
317 | |||
318 | /* get output buffer */ |
||
319 | frame->nb_samples = s->frame_len; |
||
320 | if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) |
||
321 | return ret; |
||
322 | |||
323 | if (decode_block(s, (float **)frame->extended_data, |
||
324 | avctx->codec->id == AV_CODEC_ID_BINKAUDIO_DCT)) { |
||
325 | av_log(avctx, AV_LOG_ERROR, "Incomplete packet\n"); |
||
326 | return AVERROR_INVALIDDATA; |
||
327 | } |
||
328 | get_bits_align32(gb); |
||
329 | |||
330 | frame->nb_samples = s->block_size / avctx->channels; |
||
331 | *got_frame_ptr = 1; |
||
332 | |||
333 | return consumed; |
||
334 | } |
||
335 | |||
336 | AVCodec ff_binkaudio_rdft_decoder = { |
||
337 | .name = "binkaudio_rdft", |
||
338 | .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (RDFT)"), |
||
339 | .type = AVMEDIA_TYPE_AUDIO, |
||
340 | .id = AV_CODEC_ID_BINKAUDIO_RDFT, |
||
341 | .priv_data_size = sizeof(BinkAudioContext), |
||
342 | .init = decode_init, |
||
343 | .close = decode_end, |
||
344 | .decode = decode_frame, |
||
345 | .capabilities = CODEC_CAP_DELAY | CODEC_CAP_DR1, |
||
346 | }; |
||
347 | |||
348 | AVCodec ff_binkaudio_dct_decoder = { |
||
349 | .name = "binkaudio_dct", |
||
350 | .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (DCT)"), |
||
351 | .type = AVMEDIA_TYPE_AUDIO, |
||
352 | .id = AV_CODEC_ID_BINKAUDIO_DCT, |
||
353 | .priv_data_size = sizeof(BinkAudioContext), |
||
354 | .init = decode_init, |
||
355 | .close = decode_end, |
||
356 | .decode = decode_frame, |
||
357 | .capabilities = CODEC_CAP_DELAY | CODEC_CAP_DR1, |
||
358 | };>>>>>>>>>>>>>=>>>><>>>>>>>>><> |