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 
34 #define BITSTREAM_READER_LE
35 #include "avcodec.h"
36 #include "dct.h"
37 #include "decode.h"
38 #include "get_bits.h"
39 #include "internal.h"
40 #include "rdft.h"
41 #include "wma_freqs.h"
42 
43 #define MAX_CHANNELS 2
44 #define BINK_BLOCK_MAX_SIZE (MAX_CHANNELS << 11)
45 
46 typedef struct BinkAudioContext {
48  int version_b; ///< Bink version 'b'
49  int first;
50  int channels;
51  int frame_len; ///< transform size (samples)
52  int overlap_len; ///< overlap size (samples)
54  int num_bands;
55  float root;
56  unsigned int bands[26];
57  float previous[MAX_CHANNELS][BINK_BLOCK_MAX_SIZE / 16]; ///< coeffs from previous audio block
58  float quant_table[96];
60  union {
63  } trans;
65 
66 
68 {
69  BinkAudioContext *s = avctx->priv_data;
70  int sample_rate = avctx->sample_rate;
71  int sample_rate_half;
72  int i;
73  int frame_len_bits;
74 
75  /* determine frame length */
76  if (avctx->sample_rate < 22050) {
77  frame_len_bits = 9;
78  } else if (avctx->sample_rate < 44100) {
79  frame_len_bits = 10;
80  } else {
81  frame_len_bits = 11;
82  }
83 
84  if (avctx->channels < 1 || avctx->channels > MAX_CHANNELS) {
85  av_log(avctx, AV_LOG_ERROR, "invalid number of channels: %d\n", avctx->channels);
86  return AVERROR_INVALIDDATA;
87  }
88  avctx->channel_layout = avctx->channels == 1 ? AV_CH_LAYOUT_MONO :
90 
91  s->version_b = avctx->extradata_size >= 4 && avctx->extradata[3] == 'b';
92 
93  if (avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT) {
94  // audio is already interleaved for the RDFT format variant
96  if (sample_rate > INT_MAX / avctx->channels)
97  return AVERROR_INVALIDDATA;
98  sample_rate *= avctx->channels;
99  s->channels = 1;
100  if (!s->version_b)
101  frame_len_bits += av_log2(avctx->channels);
102  } else {
103  s->channels = avctx->channels;
105  }
106 
107  s->frame_len = 1 << frame_len_bits;
108  s->overlap_len = s->frame_len / 16;
109  s->block_size = (s->frame_len - s->overlap_len) * s->channels;
110  sample_rate_half = (sample_rate + 1LL) / 2;
111  if (avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT)
112  s->root = 2.0 / (sqrt(s->frame_len) * 32768.0);
113  else
114  s->root = s->frame_len / (sqrt(s->frame_len) * 32768.0);
115  for (i = 0; i < 96; i++) {
116  /* constant is result of 0.066399999/log10(M_E) */
117  s->quant_table[i] = expf(i * 0.15289164787221953823f) * s->root;
118  }
119 
120  /* calculate number of bands */
121  for (s->num_bands = 1; s->num_bands < 25; s->num_bands++)
122  if (sample_rate_half <= ff_wma_critical_freqs[s->num_bands - 1])
123  break;
124 
125  /* populate bands data */
126  s->bands[0] = 2;
127  for (i = 1; i < s->num_bands; i++)
128  s->bands[i] = (ff_wma_critical_freqs[i - 1] * s->frame_len / sample_rate_half) & ~1;
129  s->bands[s->num_bands] = s->frame_len;
130 
131  s->first = 1;
132 
133  if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT)
134  ff_rdft_init(&s->trans.rdft, frame_len_bits, DFT_C2R);
135  else if (CONFIG_BINKAUDIO_DCT_DECODER)
136  ff_dct_init(&s->trans.dct, frame_len_bits, DCT_III);
137  else
138  av_assert0(0);
139 
140  s->pkt = av_packet_alloc();
141  if (!s->pkt)
142  return AVERROR(ENOMEM);
143 
144  return 0;
145 }
146 
147 static float get_float(GetBitContext *gb)
148 {
149  int power = get_bits(gb, 5);
150  float f = ldexpf(get_bits(gb, 23), power - 23);
151  if (get_bits1(gb))
152  f = -f;
153  return f;
154 }
155 
156 static const uint8_t rle_length_tab[16] = {
157  2, 3, 4, 5, 6, 8, 9, 10, 11, 12, 13, 14, 15, 16, 32, 64
158 };
159 
160 /**
161  * Decode Bink Audio block
162  * @param[out] out Output buffer (must contain s->block_size elements)
163  * @return 0 on success, negative error code on failure
164  */
165 static int decode_block(BinkAudioContext *s, float **out, int use_dct)
166 {
167  int ch, i, j, k;
168  float q, quant[25];
169  int width, coeff;
170  GetBitContext *gb = &s->gb;
171 
172  if (use_dct)
173  skip_bits(gb, 2);
174 
175  for (ch = 0; ch < s->channels; ch++) {
176  FFTSample *coeffs = out[ch];
177 
178  if (s->version_b) {
179  if (get_bits_left(gb) < 64)
180  return AVERROR_INVALIDDATA;
181  coeffs[0] = av_int2float(get_bits_long(gb, 32)) * s->root;
182  coeffs[1] = av_int2float(get_bits_long(gb, 32)) * s->root;
183  } else {
184  if (get_bits_left(gb) < 58)
185  return AVERROR_INVALIDDATA;
186  coeffs[0] = get_float(gb) * s->root;
187  coeffs[1] = get_float(gb) * s->root;
188  }
189 
190  if (get_bits_left(gb) < s->num_bands * 8)
191  return AVERROR_INVALIDDATA;
192  for (i = 0; i < s->num_bands; i++) {
193  int value = get_bits(gb, 8);
194  quant[i] = s->quant_table[FFMIN(value, 95)];
195  }
196 
197  k = 0;
198  q = quant[0];
199 
200  // parse coefficients
201  i = 2;
202  while (i < s->frame_len) {
203  if (s->version_b) {
204  j = i + 16;
205  } else {
206  int v = get_bits1(gb);
207  if (v) {
208  v = get_bits(gb, 4);
209  j = i + rle_length_tab[v] * 8;
210  } else {
211  j = i + 8;
212  }
213  }
214 
215  j = FFMIN(j, s->frame_len);
216 
217  width = get_bits(gb, 4);
218  if (width == 0) {
219  memset(coeffs + i, 0, (j - i) * sizeof(*coeffs));
220  i = j;
221  while (s->bands[k] < i)
222  q = quant[k++];
223  } else {
224  while (i < j) {
225  if (s->bands[k] == i)
226  q = quant[k++];
227  coeff = get_bits(gb, width);
228  if (coeff) {
229  int v;
230  v = get_bits1(gb);
231  if (v)
232  coeffs[i] = -q * coeff;
233  else
234  coeffs[i] = q * coeff;
235  } else {
236  coeffs[i] = 0.0f;
237  }
238  i++;
239  }
240  }
241  }
242 
243  if (CONFIG_BINKAUDIO_DCT_DECODER && use_dct) {
244  coeffs[0] /= 0.5;
245  s->trans.dct.dct_calc(&s->trans.dct, coeffs);
246  }
247  else if (CONFIG_BINKAUDIO_RDFT_DECODER)
248  s->trans.rdft.rdft_calc(&s->trans.rdft, coeffs);
249  }
250 
251  for (ch = 0; ch < s->channels; ch++) {
252  int j;
253  int count = s->overlap_len * s->channels;
254  if (!s->first) {
255  j = ch;
256  for (i = 0; i < s->overlap_len; i++, j += s->channels)
257  out[ch][i] = (s->previous[ch][i] * (count - j) +
258  out[ch][i] * j) / count;
259  }
260  memcpy(s->previous[ch], &out[ch][s->frame_len - s->overlap_len],
261  s->overlap_len * sizeof(*s->previous[ch]));
262  }
263 
264  s->first = 0;
265 
266  return 0;
267 }
268 
270 {
271  BinkAudioContext * s = avctx->priv_data;
272  if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT)
273  ff_rdft_end(&s->trans.rdft);
274  else if (CONFIG_BINKAUDIO_DCT_DECODER)
275  ff_dct_end(&s->trans.dct);
276 
277  av_packet_free(&s->pkt);
278 
279  return 0;
280 }
281 
283 {
284  int n = (-get_bits_count(s)) & 31;
285  if (n) skip_bits(s, n);
286 }
287 
289 {
290  BinkAudioContext *s = avctx->priv_data;
291  GetBitContext *gb = &s->gb;
292  int ret;
293 
294  if (!s->pkt->data) {
295  ret = ff_decode_get_packet(avctx, s->pkt);
296  if (ret < 0)
297  return ret;
298 
299  if (s->pkt->size < 4) {
300  av_log(avctx, AV_LOG_ERROR, "Packet is too small\n");
301  ret = AVERROR_INVALIDDATA;
302  goto fail;
303  }
304 
305  ret = init_get_bits8(gb, s->pkt->data, s->pkt->size);
306  if (ret < 0)
307  goto fail;
308 
309  /* skip reported size */
310  skip_bits_long(gb, 32);
311  }
312 
313  /* get output buffer */
314  frame->nb_samples = s->frame_len;
315  if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
316  return ret;
317 
318  if (decode_block(s, (float **)frame->extended_data,
319  avctx->codec->id == AV_CODEC_ID_BINKAUDIO_DCT)) {
320  av_log(avctx, AV_LOG_ERROR, "Incomplete packet\n");
321  return AVERROR_INVALIDDATA;
322  }
323  get_bits_align32(gb);
324  if (!get_bits_left(gb)) {
325  memset(gb, 0, sizeof(*gb));
326  av_packet_unref(s->pkt);
327  }
328 
329  frame->nb_samples = s->block_size / avctx->channels;
330 
331  return 0;
332 fail:
333  av_packet_unref(s->pkt);
334  return ret;
335 }
336 
338  .name = "binkaudio_rdft",
339  .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (RDFT)"),
340  .type = AVMEDIA_TYPE_AUDIO,
342  .priv_data_size = sizeof(BinkAudioContext),
343  .init = decode_init,
344  .close = decode_end,
346  .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_DR1,
347 };
348 
350  .name = "binkaudio_dct",
351  .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (DCT)"),
352  .type = AVMEDIA_TYPE_AUDIO,
354  .priv_data_size = sizeof(BinkAudioContext),
355  .init = decode_init,
356  .close = decode_end,
358  .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_DR1,
359 };
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:535
static float get_float(GetBitContext *gb)
Definition: binkaudio.c:147
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
Definition: error.h:59
#define MAX_CHANNELS
Definition: binkaudio.c:43
This structure describes decoded (raw) audio or video data.
Definition: frame.h:308
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:269
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:156
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:56
#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:1194
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:248
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:627
union BinkAudioContext::@24 trans
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:288
bitstream reader API header.
float previous[MAX_CHANNELS][BINK_BLOCK_MAX_SIZE/16]
coeffs from previous audio block
Definition: binkaudio.c:57
#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:44
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:153
static void get_bits_align32(GetBitContext *s)
Definition: binkaudio.c:282
static int decode_block(BinkAudioContext *s, float **out, int use_dct)
Decode Bink Audio block.
Definition: binkaudio.c:165
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:123
uint64_t channel_layout
Audio channel layout.
Definition: avcodec.h:1237
GetBitContext gb
Definition: binkaudio.c:47
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:62
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:67
void(* rdft_calc)(struct RDFTContext *s, FFTSample *z)
Definition: rdft.h:38
AVCodec ff_binkaudio_rdft_decoder
Definition: binkaudio.c:337
float quant_table[96]
Definition: binkaudio.c:58
int overlap_len
overlap size (samples)
Definition: binkaudio.c:52
sample_rate
Libavcodec external API header.
int sample_rate
samples per second
Definition: avcodec.h:1186
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:349
main external API structure.
Definition: avcodec.h:526
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:1879
#define ldexpf(x, exp)
Definition: libm.h:389
int extradata_size
Definition: avcodec.h:628
static unsigned int get_bits1(GetBitContext *s)
Definition: get_bits.h:498
AVPacket * pkt
Definition: binkaudio.c:59
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:51
int version_b
Bink version &#39;b&#39;.
Definition: binkaudio.c:48
common internal api header.
RDFTContext rdft
Definition: binkaudio.c:61
void * priv_data
Definition: avcodec.h:553
int channels
number of audio channels
Definition: avcodec.h:1187
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:72
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:355
#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
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:374
#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