1 /*
2 * ATRAC1 compatible decoder
3 * Copyright (c) 2009 Maxim Poliakovski
4 * Copyright (c) 2009 Benjamin Larsson
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 * ATRAC1 compatible decoder.
26 * This decoder handles raw ATRAC1 data and probably SDDS data.
27 */
28
29 /* Many thanks to Tim Craig for all the help! */
30
31 #include <math.h>
32 #include <stddef.h>
33 #include <stdio.h>
34
35 #include "libavutil/float_dsp.h"
36 #include "libavutil/mem_internal.h"
37
38 #include "avcodec.h"
39 #include "codec_internal.h"
40 #include "get_bits.h"
41 #include "fft.h"
42 #include "internal.h"
43 #include "sinewin.h"
44
45 #include "atrac.h"
46 #include "atrac1data.h"
47
48 #define AT1_MAX_BFU 52 ///< max number of block floating units in a sound unit
49 #define AT1_SU_SIZE 212 ///< number of bytes in a sound unit
50 #define AT1_SU_SAMPLES 512 ///< number of samples in a sound unit
51 #define AT1_FRAME_SIZE AT1_SU_SIZE * 2
52 #define AT1_SU_MAX_BITS AT1_SU_SIZE * 8
53 #define AT1_MAX_CHANNELS 2
54
55 #define AT1_QMF_BANDS 3
56 #define IDX_LOW_BAND 0
57 #define IDX_MID_BAND 1
58 #define IDX_HIGH_BAND 2
59
60 /**
61 * Sound unit struct, one unit is used per channel
62 */
63 typedef struct AT1SUCtx {
64 int log2_block_count[AT1_QMF_BANDS]; ///< log2 number of blocks in a band
65 int num_bfus; ///< number of Block Floating Units
66 float* spectrum[2];
67 DECLARE_ALIGNED(32, float, spec1)[AT1_SU_SAMPLES]; ///< mdct buffer
68 DECLARE_ALIGNED(32, float, spec2)[AT1_SU_SAMPLES]; ///< mdct buffer
69 DECLARE_ALIGNED(32, float, fst_qmf_delay)[46]; ///< delay line for the 1st stacked QMF filter
70 DECLARE_ALIGNED(32, float, snd_qmf_delay)[46]; ///< delay line for the 2nd stacked QMF filter
71 DECLARE_ALIGNED(32, float, last_qmf_delay)[256+39]; ///< delay line for the last stacked QMF filter
72 } AT1SUCtx;
73
74 /**
75 * The atrac1 context, holds all needed parameters for decoding
76 */
77 typedef struct AT1Ctx {
78 AT1SUCtx SUs[AT1_MAX_CHANNELS]; ///< channel sound unit
79 DECLARE_ALIGNED(32, float, spec)[AT1_SU_SAMPLES]; ///< the mdct spectrum buffer
80
81 DECLARE_ALIGNED(32, float, low)[256];
82 DECLARE_ALIGNED(32, float, mid)[256];
83 DECLARE_ALIGNED(32, float, high)[512];
84 float* bands[3];
85 FFTContext mdct_ctx[3];
86 void (*vector_fmul_window)(float *dst, const float *src0,
87 const float *src1, const float *win, int len);
88 } AT1Ctx;
89
90 /** size of the transform in samples in the long mode for each QMF band */
91 static const uint16_t samples_per_band[3] = {128, 128, 256};
92 static const uint8_t mdct_long_nbits[3] = {7, 7, 8};
93
94
at1_imdct(AT1Ctx * q,float * spec,float * out,int nbits,int rev_spec)95 static void at1_imdct(AT1Ctx *q, float *spec, float *out, int nbits,
96 int rev_spec)
97 {
98 FFTContext* mdct_context = &q->mdct_ctx[nbits - 5 - (nbits > 6)];
99 int transf_size = 1 << nbits;
100
101 if (rev_spec) {
102 int i;
103 for (i = 0; i < transf_size / 2; i++)
104 FFSWAP(float, spec[i], spec[transf_size - 1 - i]);
105 }
106 mdct_context->imdct_half(mdct_context, out, spec);
107 }
108
109
at1_imdct_block(AT1SUCtx * su,AT1Ctx * q)110 static int at1_imdct_block(AT1SUCtx* su, AT1Ctx *q)
111 {
112 int band_num, band_samples, log2_block_count, nbits, num_blocks, block_size;
113 unsigned int start_pos, ref_pos = 0, pos = 0;
114
115 for (band_num = 0; band_num < AT1_QMF_BANDS; band_num++) {
116 float *prev_buf;
117 int j;
118
119 band_samples = samples_per_band[band_num];
120 log2_block_count = su->log2_block_count[band_num];
121
122 /* number of mdct blocks in the current QMF band: 1 - for long mode */
123 /* 4 for short mode(low/middle bands) and 8 for short mode(high band)*/
124 num_blocks = 1 << log2_block_count;
125
126 if (num_blocks == 1) {
127 /* mdct block size in samples: 128 (long mode, low & mid bands), */
128 /* 256 (long mode, high band) and 32 (short mode, all bands) */
129 block_size = band_samples >> log2_block_count;
130
131 /* calc transform size in bits according to the block_size_mode */
132 nbits = mdct_long_nbits[band_num] - log2_block_count;
133
134 if (nbits != 5 && nbits != 7 && nbits != 8)
135 return AVERROR_INVALIDDATA;
136 } else {
137 block_size = 32;
138 nbits = 5;
139 }
140
141 start_pos = 0;
142 prev_buf = &su->spectrum[1][ref_pos + band_samples - 16];
143 for (j=0; j < num_blocks; j++) {
144 at1_imdct(q, &q->spec[pos], &su->spectrum[0][ref_pos + start_pos], nbits, band_num);
145
146 /* overlap and window */
147 q->vector_fmul_window(&q->bands[band_num][start_pos], prev_buf,
148 &su->spectrum[0][ref_pos + start_pos], ff_sine_32, 16);
149
150 prev_buf = &su->spectrum[0][ref_pos+start_pos + 16];
151 start_pos += block_size;
152 pos += block_size;
153 }
154
155 if (num_blocks == 1)
156 memcpy(q->bands[band_num] + 32, &su->spectrum[0][ref_pos + 16], 240 * sizeof(float));
157
158 ref_pos += band_samples;
159 }
160
161 /* Swap buffers so the mdct overlap works */
162 FFSWAP(float*, su->spectrum[0], su->spectrum[1]);
163
164 return 0;
165 }
166
167 /**
168 * Parse the block size mode byte
169 */
170
at1_parse_bsm(GetBitContext * gb,int log2_block_cnt[AT1_QMF_BANDS])171 static int at1_parse_bsm(GetBitContext* gb, int log2_block_cnt[AT1_QMF_BANDS])
172 {
173 int log2_block_count_tmp, i;
174
175 for (i = 0; i < 2; i++) {
176 /* low and mid band */
177 log2_block_count_tmp = get_bits(gb, 2);
178 if (log2_block_count_tmp & 1)
179 return AVERROR_INVALIDDATA;
180 log2_block_cnt[i] = 2 - log2_block_count_tmp;
181 }
182
183 /* high band */
184 log2_block_count_tmp = get_bits(gb, 2);
185 if (log2_block_count_tmp != 0 && log2_block_count_tmp != 3)
186 return AVERROR_INVALIDDATA;
187 log2_block_cnt[IDX_HIGH_BAND] = 3 - log2_block_count_tmp;
188
189 skip_bits(gb, 2);
190 return 0;
191 }
192
193
at1_unpack_dequant(GetBitContext * gb,AT1SUCtx * su,float spec[AT1_SU_SAMPLES])194 static int at1_unpack_dequant(GetBitContext* gb, AT1SUCtx* su,
195 float spec[AT1_SU_SAMPLES])
196 {
197 int bits_used, band_num, bfu_num, i;
198 uint8_t idwls[AT1_MAX_BFU]; ///< the word length indexes for each BFU
199 uint8_t idsfs[AT1_MAX_BFU]; ///< the scalefactor indexes for each BFU
200
201 /* parse the info byte (2nd byte) telling how much BFUs were coded */
202 su->num_bfus = bfu_amount_tab1[get_bits(gb, 3)];
203
204 /* calc number of consumed bits:
205 num_BFUs * (idwl(4bits) + idsf(6bits)) + log2_block_count(8bits) + info_byte(8bits)
206 + info_byte_copy(8bits) + log2_block_count_copy(8bits) */
207 bits_used = su->num_bfus * 10 + 32 +
208 bfu_amount_tab2[get_bits(gb, 2)] +
209 (bfu_amount_tab3[get_bits(gb, 3)] << 1);
210
211 /* get word length index (idwl) for each BFU */
212 for (i = 0; i < su->num_bfus; i++)
213 idwls[i] = get_bits(gb, 4);
214
215 /* get scalefactor index (idsf) for each BFU */
216 for (i = 0; i < su->num_bfus; i++)
217 idsfs[i] = get_bits(gb, 6);
218
219 /* zero idwl/idsf for empty BFUs */
220 for (i = su->num_bfus; i < AT1_MAX_BFU; i++)
221 idwls[i] = idsfs[i] = 0;
222
223 /* read in the spectral data and reconstruct MDCT spectrum of this channel */
224 for (band_num = 0; band_num < AT1_QMF_BANDS; band_num++) {
225 for (bfu_num = bfu_bands_t[band_num]; bfu_num < bfu_bands_t[band_num+1]; bfu_num++) {
226 int pos;
227
228 int num_specs = specs_per_bfu[bfu_num];
229 int word_len = !!idwls[bfu_num] + idwls[bfu_num];
230 float scale_factor = ff_atrac_sf_table[idsfs[bfu_num]];
231 bits_used += word_len * num_specs; /* add number of bits consumed by current BFU */
232
233 /* check for bitstream overflow */
234 if (bits_used > AT1_SU_MAX_BITS)
235 return AVERROR_INVALIDDATA;
236
237 /* get the position of the 1st spec according to the block size mode */
238 pos = su->log2_block_count[band_num] ? bfu_start_short[bfu_num] : bfu_start_long[bfu_num];
239
240 if (word_len) {
241 float max_quant = 1.0 / (float)((1 << (word_len - 1)) - 1);
242
243 for (i = 0; i < num_specs; i++) {
244 /* read in a quantized spec and convert it to
245 * signed int and then inverse quantization
246 */
247 spec[pos+i] = get_sbits(gb, word_len) * scale_factor * max_quant;
248 }
249 } else { /* word_len = 0 -> empty BFU, zero all specs in the empty BFU */
250 memset(&spec[pos], 0, num_specs * sizeof(float));
251 }
252 }
253 }
254
255 return 0;
256 }
257
258
at1_subband_synthesis(AT1Ctx * q,AT1SUCtx * su,float * pOut)259 static void at1_subband_synthesis(AT1Ctx *q, AT1SUCtx* su, float *pOut)
260 {
261 float temp[256];
262 float iqmf_temp[512 + 46];
263
264 /* combine low and middle bands */
265 ff_atrac_iqmf(q->bands[0], q->bands[1], 128, temp, su->fst_qmf_delay, iqmf_temp);
266
267 /* delay the signal of the high band by 39 samples */
268 memcpy( su->last_qmf_delay, &su->last_qmf_delay[256], sizeof(float) * 39);
269 memcpy(&su->last_qmf_delay[39], q->bands[2], sizeof(float) * 256);
270
271 /* combine (low + middle) and high bands */
272 ff_atrac_iqmf(temp, su->last_qmf_delay, 256, pOut, su->snd_qmf_delay, iqmf_temp);
273 }
274
275
atrac1_decode_frame(AVCodecContext * avctx,AVFrame * frame,int * got_frame_ptr,AVPacket * avpkt)276 static int atrac1_decode_frame(AVCodecContext *avctx, AVFrame *frame,
277 int *got_frame_ptr, AVPacket *avpkt)
278 {
279 const uint8_t *buf = avpkt->data;
280 int buf_size = avpkt->size;
281 AT1Ctx *q = avctx->priv_data;
282 int channels = avctx->ch_layout.nb_channels;
283 int ch, ret;
284 GetBitContext gb;
285
286
287 if (buf_size < 212 * channels) {
288 av_log(avctx, AV_LOG_ERROR, "Not enough data to decode!\n");
289 return AVERROR_INVALIDDATA;
290 }
291
292 /* get output buffer */
293 frame->nb_samples = AT1_SU_SAMPLES;
294 if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
295 return ret;
296
297 for (ch = 0; ch < channels; ch++) {
298 AT1SUCtx* su = &q->SUs[ch];
299
300 init_get_bits(&gb, &buf[212 * ch], 212 * 8);
301
302 /* parse block_size_mode, 1st byte */
303 ret = at1_parse_bsm(&gb, su->log2_block_count);
304 if (ret < 0)
305 return ret;
306
307 ret = at1_unpack_dequant(&gb, su, q->spec);
308 if (ret < 0)
309 return ret;
310
311 ret = at1_imdct_block(su, q);
312 if (ret < 0)
313 return ret;
314 at1_subband_synthesis(q, su, (float *)frame->extended_data[ch]);
315 }
316
317 *got_frame_ptr = 1;
318
319 return avctx->block_align;
320 }
321
322
atrac1_decode_end(AVCodecContext * avctx)323 static av_cold int atrac1_decode_end(AVCodecContext * avctx)
324 {
325 AT1Ctx *q = avctx->priv_data;
326
327 ff_mdct_end(&q->mdct_ctx[0]);
328 ff_mdct_end(&q->mdct_ctx[1]);
329 ff_mdct_end(&q->mdct_ctx[2]);
330
331 return 0;
332 }
333
334
atrac1_decode_init(AVCodecContext * avctx)335 static av_cold int atrac1_decode_init(AVCodecContext *avctx)
336 {
337 AT1Ctx *q = avctx->priv_data;
338 AVFloatDSPContext *fdsp;
339 int channels = avctx->ch_layout.nb_channels;
340 int ret;
341
342 avctx->sample_fmt = AV_SAMPLE_FMT_FLTP;
343
344 if (channels < 1 || channels > AT1_MAX_CHANNELS) {
345 av_log(avctx, AV_LOG_ERROR, "Unsupported number of channels: %d\n",
346 channels);
347 return AVERROR(EINVAL);
348 }
349
350 if (avctx->block_align <= 0) {
351 av_log(avctx, AV_LOG_ERROR, "Unsupported block align.");
352 return AVERROR_PATCHWELCOME;
353 }
354
355 /* Init the mdct transforms */
356 if ((ret = ff_mdct_init(&q->mdct_ctx[0], 6, 1, -1.0/ (1 << 15))) ||
357 (ret = ff_mdct_init(&q->mdct_ctx[1], 8, 1, -1.0/ (1 << 15))) ||
358 (ret = ff_mdct_init(&q->mdct_ctx[2], 9, 1, -1.0/ (1 << 15)))) {
359 av_log(avctx, AV_LOG_ERROR, "Error initializing MDCT\n");
360 return ret;
361 }
362
363 ff_init_ff_sine_windows(5);
364
365 ff_atrac_generate_tables();
366
367 fdsp = avpriv_float_dsp_alloc(avctx->flags & AV_CODEC_FLAG_BITEXACT);
368 if (!fdsp)
369 return AVERROR(ENOMEM);
370 q->vector_fmul_window = fdsp->vector_fmul_window;
371 av_free(fdsp);
372
373 q->bands[0] = q->low;
374 q->bands[1] = q->mid;
375 q->bands[2] = q->high;
376
377 /* Prepare the mdct overlap buffers */
378 q->SUs[0].spectrum[0] = q->SUs[0].spec1;
379 q->SUs[0].spectrum[1] = q->SUs[0].spec2;
380 q->SUs[1].spectrum[0] = q->SUs[1].spec1;
381 q->SUs[1].spectrum[1] = q->SUs[1].spec2;
382
383 return 0;
384 }
385
386
387 const FFCodec ff_atrac1_decoder = {
388 .p.name = "atrac1",
389 .p.long_name = NULL_IF_CONFIG_SMALL("ATRAC1 (Adaptive TRansform Acoustic Coding)"),
390 .p.type = AVMEDIA_TYPE_AUDIO,
391 .p.id = AV_CODEC_ID_ATRAC1,
392 .priv_data_size = sizeof(AT1Ctx),
393 .init = atrac1_decode_init,
394 .close = atrac1_decode_end,
395 FF_CODEC_DECODE_CB(atrac1_decode_frame),
396 .p.capabilities = AV_CODEC_CAP_DR1,
397 .p.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
398 AV_SAMPLE_FMT_NONE },
399 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE | FF_CODEC_CAP_INIT_CLEANUP,
400 };
401