FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
libvorbisenc.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2002 Mark Hills <mark@pogo.org.uk>
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * FFmpeg is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with FFmpeg; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #include <vorbis/vorbisenc.h>
22 
23 #include "libavutil/avassert.h"
24 #include "libavutil/fifo.h"
25 #include "libavutil/opt.h"
26 #include "avcodec.h"
27 #include "audio_frame_queue.h"
28 #include "internal.h"
29 #include "vorbis.h"
30 #include "vorbis_parser.h"
31 
32 
33 /* Number of samples the user should send in each call.
34  * This value is used because it is the LCD of all possible frame sizes, so
35  * an output packet will always start at the same point as one of the input
36  * packets.
37  */
38 #define OGGVORBIS_FRAME_SIZE 64
39 
40 #define BUFFER_SIZE (1024 * 64)
41 
42 typedef struct OggVorbisEncContext {
43  AVClass *av_class; /**< class for AVOptions */
44  vorbis_info vi; /**< vorbis_info used during init */
45  vorbis_dsp_state vd; /**< DSP state used for analysis */
46  vorbis_block vb; /**< vorbis_block used for analysis */
47  AVFifoBuffer *pkt_fifo; /**< output packet buffer */
48  int eof; /**< end-of-file flag */
49  int dsp_initialized; /**< vd has been initialized */
50  vorbis_comment vc; /**< VorbisComment info */
51  double iblock; /**< impulse block bias option */
52  VorbisParseContext vp; /**< parse context to get durations */
53  AudioFrameQueue afq; /**< frame queue for timestamps */
55 
56 static const AVOption options[] = {
57  { "iblock", "Sets the impulse block bias", offsetof(OggVorbisEncContext, iblock), AV_OPT_TYPE_DOUBLE, { .dbl = 0 }, -15, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM },
58  { NULL }
59 };
60 
61 static const AVCodecDefault defaults[] = {
62  { "b", "0" },
63  { NULL },
64 };
65 
66 static const AVClass vorbis_class = {
67  .class_name = "libvorbis",
68  .item_name = av_default_item_name,
69  .option = options,
70  .version = LIBAVUTIL_VERSION_INT,
71 };
72 
73 static int vorbis_error_to_averror(int ov_err)
74 {
75  switch (ov_err) {
76  case OV_EFAULT: return AVERROR_BUG;
77  case OV_EINVAL: return AVERROR(EINVAL);
78  case OV_EIMPL: return AVERROR(EINVAL);
79  default: return AVERROR_UNKNOWN;
80  }
81 }
82 
83 static av_cold int oggvorbis_init_encoder(vorbis_info *vi,
84  AVCodecContext *avctx)
85 {
87  double cfreq;
88  int ret;
89 
90  if (avctx->flags & CODEC_FLAG_QSCALE || !avctx->bit_rate) {
91  /* variable bitrate
92  * NOTE: we use the oggenc range of -1 to 10 for global_quality for
93  * user convenience, but libvorbis uses -0.1 to 1.0.
94  */
95  float q = avctx->global_quality / (float)FF_QP2LAMBDA;
96  /* default to 3 if the user did not set quality or bitrate */
97  if (!(avctx->flags & CODEC_FLAG_QSCALE))
98  q = 3.0;
99  if ((ret = vorbis_encode_setup_vbr(vi, avctx->channels,
100  avctx->sample_rate,
101  q / 10.0)))
102  goto error;
103  } else {
104  int minrate = avctx->rc_min_rate > 0 ? avctx->rc_min_rate : -1;
105  int maxrate = avctx->rc_max_rate > 0 ? avctx->rc_max_rate : -1;
106 
107  /* average bitrate */
108  if ((ret = vorbis_encode_setup_managed(vi, avctx->channels,
109  avctx->sample_rate, maxrate,
110  avctx->bit_rate, minrate)))
111  goto error;
112 
113  /* variable bitrate by estimate, disable slow rate management */
114  if (minrate == -1 && maxrate == -1)
115  if ((ret = vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL)))
116  goto error; /* should not happen */
117  }
118 
119  /* cutoff frequency */
120  if (avctx->cutoff > 0) {
121  cfreq = avctx->cutoff / 1000.0;
122  if ((ret = vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq)))
123  goto error; /* should not happen */
124  }
125 
126  /* impulse block bias */
127  if (s->iblock) {
128  if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &s->iblock)))
129  goto error;
130  }
131 
132  if (avctx->channels == 3 &&
134  avctx->channels == 4 &&
135  avctx->channel_layout != AV_CH_LAYOUT_2_2 &&
136  avctx->channel_layout != AV_CH_LAYOUT_QUAD ||
137  avctx->channels == 5 &&
140  avctx->channels == 6 &&
143  avctx->channels == 7 &&
145  avctx->channels == 8 &&
147  if (avctx->channel_layout) {
148  char name[32];
149  av_get_channel_layout_string(name, sizeof(name), avctx->channels,
150  avctx->channel_layout);
151  av_log(avctx, AV_LOG_ERROR, "%s not supported by Vorbis: "
152  "output stream will have incorrect "
153  "channel layout.\n", name);
154  } else {
155  av_log(avctx, AV_LOG_WARNING, "No channel layout specified. The encoder "
156  "will use Vorbis channel layout for "
157  "%d channels.\n", avctx->channels);
158  }
159  }
160 
161  if ((ret = vorbis_encode_setup_init(vi)))
162  goto error;
163 
164  return 0;
165 error:
166  return vorbis_error_to_averror(ret);
167 }
168 
169 /* How many bytes are needed for a buffer of length 'l' */
170 static int xiph_len(int l)
171 {
172  return 1 + l / 255 + l;
173 }
174 
176 {
177  OggVorbisEncContext *s = avctx->priv_data;
178 
179  /* notify vorbisenc this is EOF */
180  if (s->dsp_initialized)
181  vorbis_analysis_wrote(&s->vd, 0);
182 
183  vorbis_block_clear(&s->vb);
184  vorbis_dsp_clear(&s->vd);
185  vorbis_info_clear(&s->vi);
186 
188  ff_af_queue_close(&s->afq);
189  av_freep(&avctx->extradata);
190 
191  return 0;
192 }
193 
195 {
196  OggVorbisEncContext *s = avctx->priv_data;
197  ogg_packet header, header_comm, header_code;
198  uint8_t *p;
199  unsigned int offset;
200  int ret;
201 
202  vorbis_info_init(&s->vi);
203  if ((ret = oggvorbis_init_encoder(&s->vi, avctx))) {
204  av_log(avctx, AV_LOG_ERROR, "encoder setup failed\n");
205  goto error;
206  }
207  if ((ret = vorbis_analysis_init(&s->vd, &s->vi))) {
208  av_log(avctx, AV_LOG_ERROR, "analysis init failed\n");
209  ret = vorbis_error_to_averror(ret);
210  goto error;
211  }
212  s->dsp_initialized = 1;
213  if ((ret = vorbis_block_init(&s->vd, &s->vb))) {
214  av_log(avctx, AV_LOG_ERROR, "dsp init failed\n");
215  ret = vorbis_error_to_averror(ret);
216  goto error;
217  }
218 
219  vorbis_comment_init(&s->vc);
220  if (!(avctx->flags & CODEC_FLAG_BITEXACT))
221  vorbis_comment_add_tag(&s->vc, "encoder", LIBAVCODEC_IDENT);
222 
223  if ((ret = vorbis_analysis_headerout(&s->vd, &s->vc, &header, &header_comm,
224  &header_code))) {
225  ret = vorbis_error_to_averror(ret);
226  goto error;
227  }
228 
229  avctx->extradata_size = 1 + xiph_len(header.bytes) +
230  xiph_len(header_comm.bytes) +
231  header_code.bytes;
232  p = avctx->extradata = av_malloc(avctx->extradata_size +
234  if (!p) {
235  ret = AVERROR(ENOMEM);
236  goto error;
237  }
238  p[0] = 2;
239  offset = 1;
240  offset += av_xiphlacing(&p[offset], header.bytes);
241  offset += av_xiphlacing(&p[offset], header_comm.bytes);
242  memcpy(&p[offset], header.packet, header.bytes);
243  offset += header.bytes;
244  memcpy(&p[offset], header_comm.packet, header_comm.bytes);
245  offset += header_comm.bytes;
246  memcpy(&p[offset], header_code.packet, header_code.bytes);
247  offset += header_code.bytes;
248  av_assert0(offset == avctx->extradata_size);
249 
250  if ((ret = avpriv_vorbis_parse_extradata(avctx, &s->vp)) < 0) {
251  av_log(avctx, AV_LOG_ERROR, "invalid extradata\n");
252  return ret;
253  }
254 
255  vorbis_comment_clear(&s->vc);
256 
258  ff_af_queue_init(avctx, &s->afq);
259 
261  if (!s->pkt_fifo) {
262  ret = AVERROR(ENOMEM);
263  goto error;
264  }
265 
266  return 0;
267 error:
268  oggvorbis_encode_close(avctx);
269  return ret;
270 }
271 
273  const AVFrame *frame, int *got_packet_ptr)
274 {
275  OggVorbisEncContext *s = avctx->priv_data;
276  ogg_packet op;
277  int ret, duration;
278 
279  /* send samples to libvorbis */
280  if (frame) {
281  const int samples = frame->nb_samples;
282  float **buffer;
283  int c, channels = s->vi.channels;
284 
285  buffer = vorbis_analysis_buffer(&s->vd, samples);
286  for (c = 0; c < channels; c++) {
287  int co = (channels > 8) ? c :
289  memcpy(buffer[c], frame->extended_data[co],
290  samples * sizeof(*buffer[c]));
291  }
292  if ((ret = vorbis_analysis_wrote(&s->vd, samples)) < 0) {
293  av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
294  return vorbis_error_to_averror(ret);
295  }
296  if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
297  return ret;
298  } else {
299  if (!s->eof)
300  if ((ret = vorbis_analysis_wrote(&s->vd, 0)) < 0) {
301  av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
302  return vorbis_error_to_averror(ret);
303  }
304  s->eof = 1;
305  }
306 
307  /* retrieve available packets from libvorbis */
308  while ((ret = vorbis_analysis_blockout(&s->vd, &s->vb)) == 1) {
309  if ((ret = vorbis_analysis(&s->vb, NULL)) < 0)
310  break;
311  if ((ret = vorbis_bitrate_addblock(&s->vb)) < 0)
312  break;
313 
314  /* add any available packets to the output packet buffer */
315  while ((ret = vorbis_bitrate_flushpacket(&s->vd, &op)) == 1) {
316  if (av_fifo_space(s->pkt_fifo) < sizeof(ogg_packet) + op.bytes) {
317  av_log(avctx, AV_LOG_ERROR, "packet buffer is too small\n");
318  return AVERROR_BUG;
319  }
320  av_fifo_generic_write(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
321  av_fifo_generic_write(s->pkt_fifo, op.packet, op.bytes, NULL);
322  }
323  if (ret < 0) {
324  av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
325  break;
326  }
327  }
328  if (ret < 0) {
329  av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
330  return vorbis_error_to_averror(ret);
331  }
332 
333  /* check for available packets */
334  if (av_fifo_size(s->pkt_fifo) < sizeof(ogg_packet))
335  return 0;
336 
337  av_fifo_generic_read(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
338 
339  if ((ret = ff_alloc_packet2(avctx, avpkt, op.bytes)) < 0)
340  return ret;
341  av_fifo_generic_read(s->pkt_fifo, avpkt->data, op.bytes, NULL);
342 
343  avpkt->pts = ff_samples_to_time_base(avctx, op.granulepos);
344 
345  duration = avpriv_vorbis_parse_frame(&s->vp, avpkt->data, avpkt->size);
346  if (duration > 0) {
347  /* we do not know encoder delay until we get the first packet from
348  * libvorbis, so we have to update the AudioFrameQueue counts */
349  if (!avctx->delay && s->afq.frames) {
350  avctx->delay = duration;
352  s->afq.frames->duration += duration;
353  s->afq.frames->pts -= duration;
355  }
356  ff_af_queue_remove(&s->afq, duration, &avpkt->pts, &avpkt->duration);
357  }
358 
359  *got_packet_ptr = 1;
360  return 0;
361 }
362 
364  .name = "libvorbis",
365  .long_name = NULL_IF_CONFIG_SMALL("libvorbis"),
366  .type = AVMEDIA_TYPE_AUDIO,
367  .id = AV_CODEC_ID_VORBIS,
368  .priv_data_size = sizeof(OggVorbisEncContext),
370  .encode2 = oggvorbis_encode_frame,
372  .capabilities = CODEC_CAP_DELAY,
373  .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
375  .priv_class = &vorbis_class,
376  .defaults = defaults,
377 };