FFmpeg
binkaudio.c
Go to the documentation of this file.
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 
32 #include "libavutil/intfloat.h"
33 #include "libavutil/mem_internal.h"
34 
35 #define BITSTREAM_READER_LE
36 #include "avcodec.h"
37 #include "dct.h"
38 #include "decode.h"
39 #include "get_bits.h"
40 #include "internal.h"
41 #include "rdft.h"
42 #include "wma_freqs.h"
43 
44 #define MAX_CHANNELS 2
45 #define BINK_BLOCK_MAX_SIZE (MAX_CHANNELS << 11)
46 
47 typedef struct BinkAudioContext {
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)
55  int num_bands;
56  float root;
57  unsigned int bands[26];
58  float previous[MAX_CHANNELS][BINK_BLOCK_MAX_SIZE / 16]; ///< coeffs from previous audio block
59  float quant_table[96];
61  union {
64  } trans;
66 
67 
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 :
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
97  if (sample_rate > INT_MAX / avctx->channels)
98  return AVERROR_INVALIDDATA;
99  sample_rate *= avctx->channels;
100  s->channels = 1;
101  if (!s->version_b)
102  frame_len_bits += av_log2(avctx->channels);
103  } else {
104  s->channels = avctx->channels;
106  }
107 
108  s->frame_len = 1 << frame_len_bits;
109  s->overlap_len = s->frame_len / 16;
110  s->block_size = (s->frame_len - s->overlap_len) * s->channels;
111  sample_rate_half = (sample_rate + 1LL) / 2;
112  if (avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT)
113  s->root = 2.0 / (sqrt(s->frame_len) * 32768.0);
114  else
115  s->root = s->frame_len / (sqrt(s->frame_len) * 32768.0);
116  for (i = 0; i < 96; i++) {
117  /* constant is result of 0.066399999/log10(M_E) */
118  s->quant_table[i] = expf(i * 0.15289164787221953823f) * s->root;
119  }
120 
121  /* calculate number of bands */
122  for (s->num_bands = 1; s->num_bands < 25; s->num_bands++)
123  if (sample_rate_half <= ff_wma_critical_freqs[s->num_bands - 1])
124  break;
125 
126  /* populate bands data */
127  s->bands[0] = 2;
128  for (i = 1; i < s->num_bands; i++)
129  s->bands[i] = (ff_wma_critical_freqs[i - 1] * s->frame_len / sample_rate_half) & ~1;
130  s->bands[s->num_bands] = s->frame_len;
131 
132  s->first = 1;
133 
134  if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT)
135  ff_rdft_init(&s->trans.rdft, frame_len_bits, DFT_C2R);
136  else if (CONFIG_BINKAUDIO_DCT_DECODER)
137  ff_dct_init(&s->trans.dct, frame_len_bits, DCT_III);
138  else
139  av_assert0(0);
140 
141  s->pkt = av_packet_alloc();
142  if (!s->pkt)
143  return AVERROR(ENOMEM);
144 
145  return 0;
146 }
147 
148 static float get_float(GetBitContext *gb)
149 {
150  int power = get_bits(gb, 5);
151  float f = ldexpf(get_bits(gb, 23), power - 23);
152  if (get_bits1(gb))
153  f = -f;
154  return f;
155 }
156 
157 static const uint8_t rle_length_tab[16] = {
158  2, 3, 4, 5, 6, 8, 9, 10, 11, 12, 13, 14, 15, 16, 32, 64
159 };
160 
161 /**
162  * Decode Bink Audio block
163  * @param[out] out Output buffer (must contain s->block_size elements)
164  * @return 0 on success, negative error code on failure
165  */
166 static int decode_block(BinkAudioContext *s, float **out, int use_dct)
167 {
168  int ch, i, j, k;
169  float q, quant[25];
170  int width, coeff;
171  GetBitContext *gb = &s->gb;
172 
173  if (use_dct)
174  skip_bits(gb, 2);
175 
176  for (ch = 0; ch < s->channels; ch++) {
177  FFTSample *coeffs = out[ch];
178 
179  if (s->version_b) {
180  if (get_bits_left(gb) < 64)
181  return AVERROR_INVALIDDATA;
182  coeffs[0] = av_int2float(get_bits_long(gb, 32)) * s->root;
183  coeffs[1] = av_int2float(get_bits_long(gb, 32)) * s->root;
184  } else {
185  if (get_bits_left(gb) < 58)
186  return AVERROR_INVALIDDATA;
187  coeffs[0] = get_float(gb) * s->root;
188  coeffs[1] = get_float(gb) * s->root;
189  }
190 
191  if (get_bits_left(gb) < s->num_bands * 8)
192  return AVERROR_INVALIDDATA;
193  for (i = 0; i < s->num_bands; i++) {
194  int value = get_bits(gb, 8);
195  quant[i] = s->quant_table[FFMIN(value, 95)];
196  }
197 
198  k = 0;
199  q = quant[0];
200 
201  // parse coefficients
202  i = 2;
203  while (i < s->frame_len) {
204  if (s->version_b) {
205  j = i + 16;
206  } else {
207  int v = get_bits1(gb);
208  if (v) {
209  v = get_bits(gb, 4);
210  j = i + rle_length_tab[v] * 8;
211  } else {
212  j = i + 8;
213  }
214  }
215 
216  j = FFMIN(j, s->frame_len);
217 
218  width = get_bits(gb, 4);
219  if (width == 0) {
220  memset(coeffs + i, 0, (j - i) * sizeof(*coeffs));
221  i = j;
222  while (s->bands[k] < i)
223  q = quant[k++];
224  } else {
225  while (i < j) {
226  if (s->bands[k] == i)
227  q = quant[k++];
228  coeff = get_bits(gb, width);
229  if (coeff) {
230  int v;
231  v = get_bits1(gb);
232  if (v)
233  coeffs[i] = -q * coeff;
234  else
235  coeffs[i] = q * coeff;
236  } else {
237  coeffs[i] = 0.0f;
238  }
239  i++;
240  }
241  }
242  }
243 
244  if (CONFIG_BINKAUDIO_DCT_DECODER && use_dct) {
245  coeffs[0] /= 0.5;
246  s->trans.dct.dct_calc(&s->trans.dct, coeffs);
247  }
248  else if (CONFIG_BINKAUDIO_RDFT_DECODER)
249  s->trans.rdft.rdft_calc(&s->trans.rdft, coeffs);
250  }
251 
252  for (ch = 0; ch < s->channels; ch++) {
253  int j;
254  int count = s->overlap_len * s->channels;
255  if (!s->first) {
256  j = ch;
257  for (i = 0; i < s->overlap_len; i++, j += s->channels)
258  out[ch][i] = (s->previous[ch][i] * (count - j) +
259  out[ch][i] * j) / count;
260  }
261  memcpy(s->previous[ch], &out[ch][s->frame_len - s->overlap_len],
262  s->overlap_len * sizeof(*s->previous[ch]));
263  }
264 
265  s->first = 0;
266 
267  return 0;
268 }
269 
271 {
272  BinkAudioContext * s = avctx->priv_data;
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  av_packet_free(&s->pkt);
279 
280  return 0;
281 }
282 
284 {
285  int n = (-get_bits_count(s)) & 31;
286  if (n) skip_bits(s, n);
287 }
288 
290 {
291  BinkAudioContext *s = avctx->priv_data;
292  GetBitContext *gb = &s->gb;
293  int ret;
294 
295  if (!s->pkt->data) {
296  ret = ff_decode_get_packet(avctx, s->pkt);
297  if (ret < 0)
298  return ret;
299 
300  if (s->pkt->size < 4) {
301  av_log(avctx, AV_LOG_ERROR, "Packet is too small\n");
302  ret = AVERROR_INVALIDDATA;
303  goto fail;
304  }
305 
306  ret = init_get_bits8(gb, s->pkt->data, s->pkt->size);
307  if (ret < 0)
308  goto fail;
309 
310  /* skip reported size */
311  skip_bits_long(gb, 32);
312  }
313 
314  /* get output buffer */
315  frame->nb_samples = s->frame_len;
316  if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
317  return ret;
318 
319  if (decode_block(s, (float **)frame->extended_data,
320  avctx->codec->id == AV_CODEC_ID_BINKAUDIO_DCT)) {
321  av_log(avctx, AV_LOG_ERROR, "Incomplete packet\n");
322  return AVERROR_INVALIDDATA;
323  }
324  get_bits_align32(gb);
325  if (!get_bits_left(gb)) {
326  memset(gb, 0, sizeof(*gb));
327  av_packet_unref(s->pkt);
328  }
329 
330  frame->nb_samples = s->block_size / avctx->channels;
331 
332  return 0;
333 fail:
334  av_packet_unref(s->pkt);
335  return ret;
336 }
337 
339  .name = "binkaudio_rdft",
340  .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (RDFT)"),
341  .type = AVMEDIA_TYPE_AUDIO,
343  .priv_data_size = sizeof(BinkAudioContext),
344  .init = decode_init,
345  .close = decode_end,
347  .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_DR1,
348 };
349 
351  .name = "binkaudio_dct",
352  .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (DCT)"),
353  .type = AVMEDIA_TYPE_AUDIO,
355  .priv_data_size = sizeof(BinkAudioContext),
356  .init = decode_init,
357  .close = decode_end,
359  .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_DR1,
360 };
av_cold void ff_rdft_end(RDFTContext *s)
Definition: rdft.c:114
float, planar
Definition: samplefmt.h:69
const struct AVCodec * codec
Definition: avcodec.h:540
static float get_float(GetBitContext *gb)
Definition: binkaudio.c:148
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
Definition: error.h:59
#define MAX_CHANNELS
Definition: binkaudio.c:44
This structure describes decoded (raw) audio or video data.
Definition: frame.h:314
Definition: avfft.h:75
static unsigned int get_bits(GetBitContext *s, int n)
Read 1-25 bits.
Definition: get_bits.h:379
static av_cold int decode_end(AVCodecContext *avctx)
Definition: binkaudio.c:270
Definition: avfft.h:95
static void skip_bits_long(GetBitContext *s, int n)
Skips the specified number of bits.
Definition: get_bits.h:291
static av_cold int init(AVCodecContext *avctx)
Definition: avrndec.c:35
static av_always_inline float av_int2float(uint32_t i)
Reinterpret a 32-bit integer as a float.
Definition: intfloat.h:40
static const uint8_t rle_length_tab[16]
Definition: binkaudio.c:157
int size
Definition: packet.h:364
int av_log2(unsigned v)
Definition: intmath.c:26
const uint16_t ff_wma_critical_freqs[25]
Definition: wma_freqs.c:23
static CopyRet receive_frame(AVCodecContext *avctx, AVFrame *frame, int *got_frame)
Definition: crystalhd.c:560
#define AV_CH_LAYOUT_STEREO
AVCodec.
Definition: codec.h:190
#define AV_CODEC_CAP_DELAY
Encoder or decoder requires flushing with NULL input at the end in order to give the complete and cor...
Definition: codec.h:75
unsigned int bands[26]
Definition: binkaudio.c:57
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
void av_packet_free(AVPacket **pkt)
Free the packet, if the packet is reference counted, it will be unreferenced first.
Definition: avpacket.c:64
enum AVSampleFormat sample_fmt
audio sample format
Definition: avcodec.h:1199
uint8_t
#define av_cold
Definition: attributes.h:88
#define f(width, name)
Definition: cbs_vp9.c:255
int ff_decode_get_packet(AVCodecContext *avctx, AVPacket *pkt)
Called by decoders to get the next packet for decoding.
Definition: decode.c:202
GLsizei GLboolean const GLfloat * value
Definition: opengl_enc.c:108
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:632
uint8_t * data
Definition: packet.h:363
static int get_bits_count(const GetBitContext *s)
Definition: get_bits.h:219
static int binkaudio_receive_frame(AVCodecContext *avctx, AVFrame *frame)
Definition: binkaudio.c:289
bitstream reader API header.
float previous[MAX_CHANNELS][BINK_BLOCK_MAX_SIZE/16]
coeffs from previous audio block
Definition: binkaudio.c:58
#define av_log(a,...)
#define expf(x)
Definition: libm.h:283
static int get_bits_left(GetBitContext *gb)
Definition: get_bits.h:849
enum AVCodecID id
Definition: codec.h:204
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:194
#define BINK_BLOCK_MAX_SIZE
Definition: binkaudio.c:45
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:115
static void get_bits_align32(GetBitContext *s)
Definition: binkaudio.c:283
static int decode_block(BinkAudioContext *s, float **out, int use_dct)
Decode Bink Audio block.
Definition: binkaudio.c:166
const char * name
Name of the codec implementation.
Definition: codec.h:197
GLsizei count
Definition: opengl_enc.c:108
float FFTSample
Definition: avfft.h:35
#define fail()
Definition: checkasm.h:133
uint64_t channel_layout
Audio channel layout.
Definition: avcodec.h:1242
GetBitContext gb
Definition: binkaudio.c:48
audio channel layout utility functions
#define FFMIN(a, b)
Definition: common.h:96
#define width
void(* dct_calc)(struct DCTContext *s, FFTSample *data)
Definition: dct.h:38
DCTContext dct
Definition: binkaudio.c:63
Definition: dct.h:32
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several the filter must be ready for frames arriving randomly on any input any filter with several inputs will most likely require some kind of queuing mechanism It is perfectly acceptable to have a limited queue and to drop frames when the inputs are too unbalanced request_frame For filters that do not use the this method is called when a frame is wanted on an output For a it should directly call filter_frame on the corresponding output For a if there are queued frames already one of these frames should be pushed If the filter should request a frame on one of its repeatedly until at least one frame has been pushed Return or at least make progress towards producing a frame
#define s(width, name)
Definition: cbs_vp9.c:257
static av_cold int decode_init(AVCodecContext *avctx)
Definition: binkaudio.c:68
void(* rdft_calc)(struct RDFTContext *s, FFTSample *z)
Definition: rdft.h:38
AVCodec ff_binkaudio_rdft_decoder
Definition: binkaudio.c:338
float quant_table[96]
Definition: binkaudio.c:59
int overlap_len
overlap size (samples)
Definition: binkaudio.c:53
sample_rate
Libavcodec external API header.
int sample_rate
samples per second
Definition: avcodec.h:1191
static int init_get_bits8(GetBitContext *s, const uint8_t *buffer, int byte_size)
Initialize GetBitContext.
Definition: get_bits.h:677
AVCodec ff_binkaudio_dct_decoder
Definition: binkaudio.c:350
main external API structure.
Definition: avcodec.h:531
void av_packet_unref(AVPacket *pkt)
Wipe the packet.
Definition: avpacket.c:606
int ff_get_buffer(AVCodecContext *avctx, AVFrame *frame, int flags)
Get a buffer for a frame.
Definition: decode.c:1884
#define ldexpf(x, exp)
Definition: libm.h:389
int extradata_size
Definition: avcodec.h:633
static unsigned int get_bits1(GetBitContext *s)
Definition: get_bits.h:498
AVPacket * pkt
Definition: binkaudio.c:60
static void skip_bits(GetBitContext *s, int n)
Definition: get_bits.h:467
av_cold int ff_dct_init(DCTContext *s, int nbits, enum DCTTransformType inverse)
Set up DCT.
Definition: dct.c:177
static unsigned int get_bits_long(GetBitContext *s, int n)
Read 0-32 bits.
Definition: get_bits.h:546
const uint8_t * quant
int frame_len
transform size (samples)
Definition: binkaudio.c:52
int version_b
Bink version &#39;b&#39;.
Definition: binkaudio.c:49
common internal api header.
RDFTContext rdft
Definition: binkaudio.c:62
void * priv_data
Definition: avcodec.h:558
int channels
number of audio channels
Definition: avcodec.h:1192
AVPacket * av_packet_alloc(void)
Allocate an AVPacket and set its fields to default values.
Definition: avpacket.c:53
static const double coeff[2][5]
Definition: vf_owdenoise.c:73
av_cold void ff_dct_end(DCTContext *s)
Definition: dct.c:221
FILE * out
Definition: movenc.c:54
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later.That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another.Frame references ownership and permissions
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: frame.h:361
#define AV_CH_LAYOUT_MONO
av_cold int ff_rdft_init(RDFTContext *s, int nbits, enum RDFTransformType trans)
Set up a real FFT.
Definition: rdft.c:88
This structure stores compressed data.
Definition: packet.h:340
union BinkAudioContext::@23 trans
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:380
#define AV_CODEC_CAP_DR1
Codec uses get_buffer() for allocating buffers and supports custom allocators.
Definition: codec.h:50
int i
Definition: input.c:407