2 * DV format muxer/demuxer
3 * Copyright (c) 2003 Roman Shaposhnik
5 * Many thanks to Dan Dennedy <dan@dennedy.org> for providing wealth
6 * of DV technical info.
8 * This library 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 of the License, or (at your option) any later version.
13 * This library 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.
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with this library; if not, write to the Free Software
20 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
26 * The reason why the following three big ugly looking tables are
27 * here is my lack of DV spec IEC 61834. The tables were basically
28 * constructed to make code that places packs in SSYB, VAUX and
29 * AAUX blocks very simple and table-driven. They conform to the
30 * SMPTE 314M and the output of my personal DV camcorder, neither
31 * of which is sufficient for a reliable DV stream producing. Thus
32 * while code is still in development I'll be gathering input from
33 * people with different DV equipment and modifying the tables to
34 * accommodate all the quirks. Later on, if possible, some of them
35 * will be folded into smaller tables and/or switch-if logic. For
36 * now, my only excuse is -- they don't eat up that much of a space.
39 static const int dv_ssyb_packs_dist[12][6] = {
40 { 0x13, 0x13, 0x13, 0x13, 0x13, 0x13 },
41 { 0x13, 0x13, 0x13, 0x13, 0x13, 0x13 },
42 { 0x13, 0x13, 0x13, 0x13, 0x13, 0x13 },
43 { 0x13, 0x13, 0x13, 0x13, 0x13, 0x13 },
44 { 0x13, 0x13, 0x13, 0x13, 0x13, 0x13 },
45 { 0x13, 0x13, 0x13, 0x13, 0x13, 0x13 },
46 { 0x13, 0x62, 0x63, 0x13, 0x62, 0x63 },
47 { 0x13, 0x62, 0x63, 0x13, 0x62, 0x63 },
48 { 0x13, 0x62, 0x63, 0x13, 0x62, 0x63 },
49 { 0x13, 0x62, 0x63, 0x13, 0x62, 0x63 },
50 { 0x13, 0x62, 0x63, 0x13, 0x62, 0x63 },
51 { 0x13, 0x62, 0x63, 0x13, 0x62, 0x63 },
54 static const int dv_vaux_packs_dist[12][15] = {
55 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
56 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
57 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
58 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
59 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
60 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
61 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
62 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
63 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
64 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
65 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
66 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
67 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
68 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
69 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
70 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
71 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
72 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
73 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
74 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
75 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
76 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
77 { 0x60, 0x61, 0x62, 0x63, 0xff, 0xff, 0xff, 0xff, 0xff,
78 0x60, 0x61, 0x62, 0x63, 0xff, 0xff },
81 static const int dv_aaux_packs_dist[12][9] = {
82 { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff },
83 { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff },
84 { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff },
85 { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff },
86 { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff },
87 { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff },
88 { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff },
89 { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff },
90 { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff },
91 { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff },
92 { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff },
93 { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff },
96 enum dv_section_type {
97 dv_sect_header = 0x1f,
98 dv_sect_subcode = 0x3f,
100 dv_sect_audio = 0x76,
101 dv_sect_video = 0x96,
105 dv_header525 = 0x3f, /* see dv_write_pack for important details on */
106 dv_header625 = 0xbf, /* these two packs */
108 dv_audio_source = 0x50,
109 dv_audio_control = 0x51,
110 dv_audio_recdate = 0x52,
111 dv_audio_rectime = 0x53,
112 dv_video_source = 0x60,
113 dv_video_control = 0x61,
114 dv_viedo_recdate = 0x62,
115 dv_video_rectime = 0x63,
116 dv_unknown_pack = 0xff,
119 static const uint16_t dv_audio_shuffle525[10][9] = {
120 { 0, 30, 60, 20, 50, 80, 10, 40, 70 }, /* 1st channel */
121 { 6, 36, 66, 26, 56, 86, 16, 46, 76 },
122 { 12, 42, 72, 2, 32, 62, 22, 52, 82 },
123 { 18, 48, 78, 8, 38, 68, 28, 58, 88 },
124 { 24, 54, 84, 14, 44, 74, 4, 34, 64 },
126 { 1, 31, 61, 21, 51, 81, 11, 41, 71 }, /* 2nd channel */
127 { 7, 37, 67, 27, 57, 87, 17, 47, 77 },
128 { 13, 43, 73, 3, 33, 63, 23, 53, 83 },
129 { 19, 49, 79, 9, 39, 69, 29, 59, 89 },
130 { 25, 55, 85, 15, 45, 75, 5, 35, 65 },
133 static const uint16_t dv_audio_shuffle625[12][9] = {
134 { 0, 36, 72, 26, 62, 98, 16, 52, 88}, /* 1st channel */
135 { 6, 42, 78, 32, 68, 104, 22, 58, 94},
136 { 12, 48, 84, 2, 38, 74, 28, 64, 100},
137 { 18, 54, 90, 8, 44, 80, 34, 70, 106},
138 { 24, 60, 96, 14, 50, 86, 4, 40, 76},
139 { 30, 66, 102, 20, 56, 92, 10, 46, 82},
141 { 1, 37, 73, 27, 63, 99, 17, 53, 89}, /* 2nd channel */
142 { 7, 43, 79, 33, 69, 105, 23, 59, 95},
143 { 13, 49, 85, 3, 39, 75, 29, 65, 101},
144 { 19, 55, 91, 9, 45, 81, 35, 71, 107},
145 { 25, 61, 97, 15, 51, 87, 5, 41, 77},
146 { 31, 67, 103, 21, 57, 93, 11, 47, 83},
149 static const int dv_audio_frequency[3] = {
153 const DVprofile dv_profiles[2] = {
155 .frame_size = 120000, /* 525/60 system (NTSC) */
159 .frame_rate_base = 1001,
161 /* .video_place = dv_place_411, */
163 .audio_min_samples = { 1580, 1452, 1053 }, /* for 48, 44.1 and 32Khz */
164 .audio_samples_dist = { 1602, 1601, 1602, 1601, 1602 },
165 .audio_shuffle = dv_audio_shuffle525,
168 .frame_size = 144000, /* 625/50 system (PAL) */
171 .frame_rate_base = 1,
174 /* .video_place = dv_place_420, */
176 .audio_min_samples = { 1896, 1742, 1264 }, /* for 48, 44.1 and 32Khz */
177 .audio_samples_dist = { 1920, 1920, 1920, 1920, 1920 },
178 .audio_shuffle = dv_audio_shuffle525,
182 static inline uint16_t dv_audio_12to16(uint16_t sample)
184 uint16_t shift, result;
186 sample = (sample < 0x800) ? sample : sample | 0xf000;
187 shift = (sample & 0xf00) >> 8;
189 if (shift < 0x2 || shift > 0xd) {
191 } else if (shift < 0x8) {
193 result = (sample - (256 * shift)) << shift;
196 result = ((sample + ((256 * shift) + 1)) << shift) - 1;
202 static int dv_write_pack(enum dv_pack_type pack_id, DVMuxContext *c, uint8_t* buf)
208 buf[0] = (uint8_t)pack_id;
210 case dv_header525: /* I can't imagine why these two weren't defined as real */
211 case dv_header625: /* packs in SMPTE314M -- they definitely look like ones */
212 buf[1] = 0xf8 | /* reserved -- always 1 */
213 (0 & 0x07); /* APT: Track application ID */
214 buf[2] = (0 << 7) | /* TF1: audio data is 0 - valid; 1 - invalid */
215 (0x0f << 3) | /* reserved -- always 1 */
216 (0 & 0x07); /* AP1: Audio application ID */
217 buf[3] = (0 << 7) | /* TF2: video data is 0 - valid; 1 - invalid */
218 (0x0f << 3) | /* reserved -- always 1 */
219 (0 & 0x07); /* AP2: Video application ID */
220 buf[4] = (0 << 7) | /* TF3: subcode(SSYB) is 0 - valid; 1 - invalid */
221 (0x0f << 3) | /* reserved -- always 1 */
222 (0 & 0x07); /* AP3: Subcode application ID */
225 ct = (time_t)(c->frames / ((float)c->sys->frame_rate /
226 (float)c->sys->frame_rate_base));
227 localtime_r(&ct, &tc);
229 * LTC drop-frame frame counter drops two frames (0 and 1) every
230 * minute, unless it is exactly divisible by 10
232 ltc_frame = (c->frames + 2*ct/60 - 2*ct/600) % c->sys->ltc_divisor;
233 buf[1] = (0 << 7) | /* Color fame: 0 - unsync; 1 - sync mode */
234 (1 << 6) | /* Drop frame timecode: 0 - nondrop; 1 - drop */
235 ((ltc_frame / 10) << 4) | /* Tens of frames */
236 (ltc_frame % 10); /* Units of frames */
237 buf[2] = (1 << 7) | /* Biphase mark polarity correction: 0 - even; 1 - odd */
238 ((tc.tm_sec / 10) << 4) | /* Tens of seconds */
239 (tc.tm_sec % 10); /* Units of seconds */
240 buf[3] = (1 << 7) | /* Binary group flag BGF0 */
241 ((tc.tm_min / 10) << 4) | /* Tens of minutes */
242 (tc.tm_min % 10); /* Units of minutes */
243 buf[4] = (1 << 7) | /* Binary group flag BGF2 */
244 (1 << 6) | /* Binary group flag BGF1 */
245 ((tc.tm_hour / 10) << 4) | /* Tens of hours */
246 (tc.tm_hour % 10); /* Units of hours */
248 case dv_audio_source: /* AAUX source pack */
249 buf[1] = (0 << 7) | /* locked mode */
250 (1 << 6) | /* reserved -- always 1 */
251 (dv_audio_frame_size(c->sys, c->frames) -
252 c->sys->audio_min_samples[0]);
254 buf[2] = (0 << 7) | /* multi-stereo */
255 (0 << 5) | /* #of audio channels per block: 0 -- 1 channel */
256 (0 << 4) | /* pair bit: 0 -- one pair of channels */
258 buf[3] = (1 << 7) | /* res */
259 (1 << 6) | /* multi-language flag */
260 (c->sys->dsf << 5) | /* system: 60fields/50fields */
261 0; /* definition: 0 -- SD (525/625) */
262 buf[4] = (1 << 7) | /* emphasis: 1 -- off */
263 (0 << 6) | /* emphasis time constant: 0 -- reserved */
264 (0 << 3) | /* frequency: 0 -- 48Khz, 1 -- 44,1Khz, 2 -- 32Khz */
265 0; /* quantization: 0 -- 16bit linear, 1 -- 12bit nonlinear */
267 case dv_audio_control:
268 buf[1] = (0 << 6) | /* copy protection: 0 -- unrestricted */
269 (1 << 4) | /* input source: 1 -- digital input */
270 (3 << 2) | /* compression: 3 -- no information */
271 0; /* misc. info/SMPTE emphasis off */
272 buf[2] = (1 << 7) | /* recording start point: 1 -- no */
273 (1 << 6) | /* recording end point: 1 -- no */
274 (1 << 3) | /* recording mode: 1 -- original */
276 buf[3] = (1 << 7) | /* direction: 1 -- forward */
278 buf[4] = (1 << 7) | /* reserved -- always 1 */
279 0x7f; /* genre category */
281 case dv_audio_recdate:
282 case dv_viedo_recdate: /* VAUX recording date */
283 ct = c->start_time + (time_t)(c->frames /
284 ((float)c->sys->frame_rate / (float)c->sys->frame_rate_base));
285 localtime_r(&ct, &tc);
286 buf[1] = 0xff; /* ds, tm, tens of time zone, units of time zone */
287 /* 0xff is very likely to be "unknown" */
288 buf[2] = (3 << 6) | /* reserved -- always 1 */
289 ((tc.tm_mday / 10) << 4) | /* Tens of day */
290 (tc.tm_mday % 10); /* Units of day */
291 buf[3] = /* we set high 4 bits to 0, shouldn't we set them to week? */
292 (tc.tm_mon % 10); /* Units of month */
293 buf[4] = (((tc.tm_year % 100) / 10) << 4) | /* Tens of year */
294 (tc.tm_year % 10); /* Units of year */
296 case dv_audio_rectime: /* AAUX recording time */
297 case dv_video_rectime: /* VAUX recording time */
298 ct = c->start_time + (time_t)(c->frames /
299 ((float)c->sys->frame_rate / (float)c->sys->frame_rate_base));
300 localtime_r(&ct, &tc);
301 buf[1] = (3 << 6) | /* reserved -- always 1 */
302 0x3f; /* tens of frame, units of frame: 0x3f - "unknown" ? */
303 buf[2] = (1 << 7) | /* reserved -- always 1 */
304 ((tc.tm_sec / 10) << 4) | /* Tens of seconds */
305 (tc.tm_sec % 10); /* Units of seconds */
306 buf[3] = (1 << 7) | /* reserved -- always 1 */
307 ((tc.tm_min / 10) << 4) | /* Tens of minutes */
308 (tc.tm_min % 10); /* Units of minutes */
309 buf[4] = (3 << 6) | /* reserved -- always 1 */
310 ((tc.tm_hour / 10) << 4) | /* Tens of hours */
311 (tc.tm_hour % 10); /* Units of hours */
313 case dv_video_source:
314 buf[1] = 0xff; /* reserved -- always 1 */
315 buf[2] = (1 << 7) | /* B/W: 0 - b/w, 1 - color */
316 (1 << 6) | /* following CLF is valid - 0, invalid - 1 */
317 (3 << 4) | /* CLF: color frames id (see ITU-R BT.470-4) */
318 0xf; /* reserved -- always 1 */
319 buf[3] = (3 << 6) | /* reserved -- always 1 */
320 (c->sys->dsf << 5) | /* system: 60fields/50fields */
321 0; /* signal type video compression */
322 buf[4] = 0xff; /* VISC: 0xff -- no information */
324 case dv_video_control:
325 buf[1] = (0 << 6) | /* Copy generation management (CGMS) 0 -- free */
326 0x3f; /* reserved -- always 1 */
327 buf[2] = 0xc8 | /* reserved -- always b11001xxx */
329 buf[3] = (1 << 7) | /* Frame/field flag 1 -- frame, 0 -- field */
330 (1 << 6) | /* First/second field flag 0 -- field 2, 1 -- field 1 */
331 (1 << 5) | /* Frame change flag 0 -- same picture as before, 1 -- different */
332 (1 << 4) | /* 1 - interlaced, 0 - noninterlaced */
333 0xc; /* reserved -- always b1100 */
334 buf[4] = 0xff; /* reserved -- always 1 */
337 buf[1] = buf[2] = buf[3] = buf[4] = 0xff;
342 static inline int dv_write_dif_id(enum dv_section_type t, uint8_t seq_num,
343 uint8_t dif_num, uint8_t* buf)
345 buf[0] = (uint8_t)t; /* Section type */
346 buf[1] = (seq_num<<4) | /* DIF seq number 0-9 for 525/60; 0-11 for 625/50 */
347 (0 << 3) | /* FSC: for 50Mb/s 0 - first channel; 1 - second */
348 7; /* reserved -- always 1 */
349 buf[2] = dif_num; /* DIF block number Video: 0-134, Audio: 0-8 */
353 static inline int dv_write_ssyb_id(uint8_t syb_num, uint8_t fr, uint8_t* buf)
355 if (syb_num == 0 || syb_num == 6) {
356 buf[0] = (fr<<7) | /* FR ID 1 - first half of each channel; 0 - second */
357 (0<<4) | /* AP3 (Subcode application ID) */
358 0x0f; /* reserved -- always 1 */
360 else if (syb_num == 11) {
361 buf[0] = (fr<<7) | /* FR ID 1 - first half of each channel; 0 - second */
362 0x7f; /* reserved -- always 1 */
365 buf[0] = (fr<<7) | /* FR ID 1 - first half of each channel; 0 - second */
366 (0<<4) | /* APT (Track application ID) */
367 0x0f; /* reserved -- always 1 */
369 buf[1] = 0xf0 | /* reserved -- always 1 */
370 (syb_num & 0x0f); /* SSYB number 0 - 11 */
371 buf[2] = 0xff; /* reserved -- always 1 */
375 void dv_format_frame(DVMuxContext *c, uint8_t* buf)
379 for (i = 0; i < c->sys->difseg_size; i++) {
380 memset(buf, 0xff, 80 * 6); /* First 6 DIF blocks are for control data */
382 /* DV header: 1DIF */
383 buf += dv_write_dif_id(dv_sect_header, i, 0, buf);
384 buf += dv_write_pack((c->sys->dsf ? dv_header625 : dv_header525), c, buf);
385 buf += 72; /* unused bytes */
387 /* DV subcode: 2DIFs */
388 for (j = 0; j < 2; j++) {
389 buf += dv_write_dif_id( dv_sect_subcode, i, j, buf);
390 for (k = 0; k < 6; k++) {
391 buf += dv_write_ssyb_id(k, (i < c->sys->difseg_size/2), buf);
392 buf += dv_write_pack(dv_ssyb_packs_dist[i][k], c, buf);
394 buf += 29; /* unused bytes */
398 for (j = 0; j < 3; j++) {
399 buf += dv_write_dif_id(dv_sect_vaux, i, j, buf);
400 for (k = 0; k < 15 ; k++)
401 buf += dv_write_pack(dv_vaux_packs_dist[i][k], c, buf);
402 buf += 2; /* unused bytes */
405 /* DV Audio/Video: 135 Video DIFs + 9 Audio DIFs */
406 for (j = 0; j < 135; j++) {
408 buf += dv_write_dif_id(dv_sect_audio, i, j/15, buf);
409 buf += dv_write_pack(dv_aaux_packs_dist[i][j/15], c, buf);
410 buf += 72; /* shuffled PCM audio */
412 buf += dv_write_dif_id(dv_sect_video, i, j, buf);
413 buf += 77; /* 1 video macro block: 1 bytes control
414 4 * 14 bytes Y 8x8 data
416 10 bytes Cb 8x8 data */
421 void dv_inject_audio(DVMuxContext *c, const uint8_t* pcm, uint8_t* frame_ptr)
424 for (i = 0; i < c->sys->difseg_size; i++) {
425 frame_ptr += 6 * 80; /* skip DIF segment header */
426 for (j = 0; j < 9; j++) {
427 for (d = 8; d < 80; d+=2) {
428 of = c->sys->audio_shuffle[i][j] + (d - 8)/2 * c->sys->audio_stride;
429 frame_ptr[d] = pcm[of*2+1]; // FIXME: may be we have to admit
430 frame_ptr[d+1] = pcm[of*2]; // that DV is a big endian PCM
432 frame_ptr += 16 * 80; /* 15 Video DIFs + 1 Audio DIF */
437 void dv_inject_video(DVMuxContext *c, const uint8_t* video_data, uint8_t* frame_ptr)
442 for (i = 0; i < c->sys->difseg_size; i++) {
443 ptr += 6 * 80; /* skip DIF segment header */
444 for (j = 0; j < 135; j++) {
446 ptr += 80; /* skip Audio DIF */
448 memcpy(frame_ptr + ptr, video_data + ptr, 77);
454 int dv_audio_frame_size(const DVprofile* sys, int frame)
456 return sys->audio_samples_dist[frame % (sizeof(sys->audio_samples_dist)/
457 sizeof(sys->audio_samples_dist[0]))];
460 const DVprofile* dv_frame_profile(uint8_t* frame)
462 return &dv_profiles[!!(frame[3] & 0x80)]; /* Header, DSF flag */
466 * This is the dumbest implementation of all -- it simply looks at
467 * a fixed offset and if pack isn't there -- fails. We might want
468 * to have a fallback mechanism for complete search of missing packs.
470 const uint8_t* dv_extract_pack(uint8_t* frame, enum dv_pack_type t)
475 case dv_audio_source:
476 offs = (80*6 + 80*16*3 + 3);
478 case dv_audio_control:
479 offs = (80*6 + 80*16*4 + 3);
481 case dv_video_control:
482 offs = (80*5 + 48 + 5);
488 return (frame[offs] == t ? &frame[offs] : NULL);
492 * There's a couple of assumptions being made here:
493 * 1. By default we silence erroneous (0x8000/16bit 0x800/12bit) audio samples.
494 * We can pass them upwards when ffmpeg will be ready to deal with them.
495 * 2. We don't do software emphasis.
496 * 3. Audio is always returned as 16bit linear samples: 12bit nonlinear samples
497 * are converted into 16bit linear ones.
499 int dv_extract_audio(uint8_t* frame, uint8_t* pcm, AVCodecContext* avctx)
501 int size, i, j, d, of, smpls, freq, quant;
503 const DVprofile* sys;
504 const uint8_t* as_pack;
506 as_pack = dv_extract_pack(frame, dv_audio_source);
507 if (!as_pack) /* No audio ? */
510 sys = dv_frame_profile(frame);
511 smpls = as_pack[1] & 0x3f; /* samples in this frame - min. samples */
512 freq = (as_pack[4] >> 3) & 0x07; /* 0 - 48KHz, 1 - 44,1kHz, 2 - 32 kHz */
513 quant = as_pack[4] & 0x07; /* 0 - 16bit linear, 1 - 12bit nonlinear */
516 return -1; /* Unsupported quantization */
518 avctx->sample_rate = dv_audio_frequency[freq];
520 avctx->bit_rate = avctx->channels * avctx->sample_rate * 16;
522 // avctx->frame_size =
524 size = (sys->audio_min_samples[freq] + smpls) * 4; /* 2ch, 2bytes */
526 /* for each DIF segment */
527 for (i = 0; i < sys->difseg_size; i++) {
528 frame += 6 * 80; /* skip DIF segment header */
529 for (j = 0; j < 9; j++) {
530 for (d = 8; d < 80; d += 2) {
531 if (quant == 0) { /* 16bit quantization */
532 of = sys->audio_shuffle[i][j] + (d - 8)/2 * sys->audio_stride;
533 pcm[of*2] = frame[d+1]; // FIXME: may be we have to admit
534 pcm[of*2+1] = frame[d]; // that DV is a big endian PCM
535 if (pcm[of*2+1] == 0x80 && pcm[of*2] == 0x00)
537 } else { /* 12bit quantization */
538 if (i >= sys->difseg_size/2)
539 goto out; /* We're not doing 4ch at this time */
541 lc = ((uint16_t)frame[d] << 4) |
542 ((uint16_t)frame[d+2] >> 4);
543 rc = ((uint16_t)frame[d+1] << 4) |
544 ((uint16_t)frame[d+2] & 0x0f);
545 lc = (lc == 0x800 ? 0 : dv_audio_12to16(lc));
546 rc = (rc == 0x800 ? 0 : dv_audio_12to16(rc));
548 of = sys->audio_shuffle[i][j] + (d - 8)/3 * sys->audio_stride;
549 pcm[of*2] = lc & 0xff; // FIXME: may be we have to admit
550 pcm[of*2+1] = lc >> 8; // that DV is a big endian PCM
551 of = sys->audio_shuffle[i+sys->difseg_size/2][j] +
552 (d - 8)/3 * sys->audio_stride;
553 pcm[of*2] = rc & 0xff; // FIXME: may be we have to admit
554 pcm[of*2+1] = rc >> 8; // that DV is a big endian PCM
559 frame += 16 * 80; /* 15 Video DIFs + 1 Audio DIF */
567 /* FIXME: The following three functions could be underengineered ;-) */
568 void dv_assemble_frame(DVMuxContext *c, const uint8_t* video, const uint8_t* audio, int asize)
571 uint8_t* frame = &c->frame_buf[0];
574 if (c->has_audio && c->has_video) { /* must be a stale frame */
575 dv_format_frame(c, frame);
577 c->has_audio = c->has_video = 0;
581 /* FIXME: we have to have more sensible approach than this one */
583 fprintf(stderr, "Can't process DV frame #%d. Insufficient audio data or severe sync problem.\n", c->frames);
585 dv_inject_video(c, video, frame);
589 reqasize = 4 * dv_audio_frame_size(c->sys, c->frames);
590 fsize = fifo_size(&c->audio_data, c->audio_data.rptr);
591 if (fsize + asize >= reqasize) {
592 if (fsize >= reqasize) {
593 fifo_read(&c->audio_data, &pcm[0], reqasize, &c->audio_data.rptr);
595 fifo_read(&c->audio_data, &pcm[0], fsize, &c->audio_data.rptr);
596 memcpy(&pcm[fsize], &audio[0], reqasize - fsize);
597 audio += reqasize - fsize;
598 asize -= reqasize - fsize;
600 dv_inject_audio(c, &pcm[0], frame);
604 /* FIXME: we have to have more sensible approach than this one */
605 if (fifo_size(&c->audio_data, c->audio_data.rptr) + asize >= AVCODEC_MAX_AUDIO_FRAME_SIZE)
606 fprintf(stderr, "Can't process DV frame #%d. Insufficient video data or severe sync problem.\n", c->frames);
607 fifo_write(&c->audio_data, (uint8_t *)audio, asize, &c->audio_data.wptr);
611 int dv_core_init(DVMuxContext *c, AVStream *streams[])
613 /* We have to sort out where audio and where video stream is */
614 if (streams[0]->codec.codec_type == CODEC_TYPE_VIDEO &&
615 streams[1]->codec.codec_type == CODEC_TYPE_AUDIO) {
619 else if (streams[1]->codec.codec_type == CODEC_TYPE_VIDEO &&
620 streams[0]->codec.codec_type == CODEC_TYPE_AUDIO) {
626 /* Some checks -- DV format is very picky about its incoming streams */
627 if (streams[c->vst]->codec.codec_id != CODEC_ID_DVVIDEO ||
628 streams[c->ast]->codec.codec_id != CODEC_ID_PCM_S16LE)
630 if (streams[c->ast]->codec.sample_rate != 48000 ||
631 streams[c->ast]->codec.channels != 2)
634 if (streams[c->vst]->codec.frame_rate == 25 &&
635 streams[c->vst]->codec.frame_rate_base == 1) {
636 /* May be we have to pick sys for every frame */
637 c->sys = &dv_profiles[1];
639 else if (streams[c->vst]->codec.frame_rate == 30000 &&
640 streams[c->vst]->codec.frame_rate_base == 1001) {
641 /* May be we have to pick sys for every frame */
642 c->sys = &dv_profiles[0];
646 /* Ok, everything seems to be in working order */
648 c->has_audio = c->has_video = 0;
649 c->start_time = time(NULL);
650 c->aspect = 0; /* 4:3 is the default */
651 if (streams[c->vst]->codec.aspect_ratio == 16.0 / 9.0)
654 if (fifo_init(&c->audio_data, AVCODEC_MAX_AUDIO_FRAME_SIZE) < 0)
657 dv_format_frame(c, &c->frame_buf[0]);
665 void dv_core_delete(DVMuxContext *c)
667 fifo_free(&c->audio_data);