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 LIBVORBIS_FRAME_SIZE 64
39
40 #define BUFFER_SIZE (1024 * 64)
41
42 typedef struct LibvorbisEncContext {
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 AVVorbisParseContext *vp; /**< parse context to get durations */
53 AudioFrameQueue afq; /**< frame queue for timestamps */
54 } LibvorbisEncContext;
55
56 static const AVOption options[] = {
57 { "iblock", "Sets the impulse block bias", offsetof(LibvorbisEncContext, 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 const uint8_t vorbis_encoding_channel_layout_offsets[8][8] = {
74 { 0 },
75 { 0, 1 },
76 { 0, 2, 1 },
77 { 0, 1, 2, 3 },
78 { 0, 2, 1, 3, 4 },
79 { 0, 2, 1, 4, 5, 3 },
80 { 0, 2, 1, 5, 6, 4, 3 },
81 { 0, 2, 1, 6, 7, 4, 5, 3 },
82 };
83
vorbis_error_to_averror(int ov_err)84 static int vorbis_error_to_averror(int ov_err)
85 {
86 switch (ov_err) {
87 case OV_EFAULT: return AVERROR_BUG;
88 case OV_EINVAL: return AVERROR(EINVAL);
89 case OV_EIMPL: return AVERROR(EINVAL);
90 default: return AVERROR_UNKNOWN;
91 }
92 }
93
libvorbis_setup(vorbis_info * vi,AVCodecContext * avctx)94 static av_cold int libvorbis_setup(vorbis_info *vi, AVCodecContext *avctx)
95 {
96 LibvorbisEncContext *s = avctx->priv_data;
97 double cfreq;
98 int ret;
99
100 if (avctx->flags & AV_CODEC_FLAG_QSCALE || !avctx->bit_rate) {
101 /* variable bitrate
102 * NOTE: we use the oggenc range of -1 to 10 for global_quality for
103 * user convenience, but libvorbis uses -0.1 to 1.0.
104 */
105 float q = avctx->global_quality / (float)FF_QP2LAMBDA;
106 /* default to 3 if the user did not set quality or bitrate */
107 if (!(avctx->flags & AV_CODEC_FLAG_QSCALE))
108 q = 3.0;
109 if ((ret = vorbis_encode_setup_vbr(vi, avctx->channels,
110 avctx->sample_rate,
111 q / 10.0)))
112 goto error;
113 } else {
114 int minrate = avctx->rc_min_rate > 0 ? avctx->rc_min_rate : -1;
115 int maxrate = avctx->rc_max_rate > 0 ? avctx->rc_max_rate : -1;
116
117 /* average bitrate */
118 if ((ret = vorbis_encode_setup_managed(vi, avctx->channels,
119 avctx->sample_rate, maxrate,
120 avctx->bit_rate, minrate)))
121 goto error;
122
123 /* variable bitrate by estimate, disable slow rate management */
124 if (minrate == -1 && maxrate == -1)
125 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL)))
126 goto error; /* should not happen */
127 }
128
129 /* cutoff frequency */
130 if (avctx->cutoff > 0) {
131 cfreq = avctx->cutoff / 1000.0;
132 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq)))
133 goto error; /* should not happen */
134 }
135
136 /* impulse block bias */
137 if (s->iblock) {
138 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &s->iblock)))
139 goto error;
140 }
141
142 if (avctx->channels == 3 &&
143 avctx->channel_layout != (AV_CH_LAYOUT_STEREO|AV_CH_FRONT_CENTER) ||
144 avctx->channels == 4 &&
145 avctx->channel_layout != AV_CH_LAYOUT_2_2 &&
146 avctx->channel_layout != AV_CH_LAYOUT_QUAD ||
147 avctx->channels == 5 &&
148 avctx->channel_layout != AV_CH_LAYOUT_5POINT0 &&
149 avctx->channel_layout != AV_CH_LAYOUT_5POINT0_BACK ||
150 avctx->channels == 6 &&
151 avctx->channel_layout != AV_CH_LAYOUT_5POINT1 &&
152 avctx->channel_layout != AV_CH_LAYOUT_5POINT1_BACK ||
153 avctx->channels == 7 &&
154 avctx->channel_layout != (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_CENTER) ||
155 avctx->channels == 8 &&
156 avctx->channel_layout != AV_CH_LAYOUT_7POINT1) {
157 if (avctx->channel_layout) {
158 char name[32];
159 av_get_channel_layout_string(name, sizeof(name), avctx->channels,
160 avctx->channel_layout);
161 av_log(avctx, AV_LOG_ERROR, "%s not supported by Vorbis: "
162 "output stream will have incorrect "
163 "channel layout.\n", name);
164 } else {
165 av_log(avctx, AV_LOG_WARNING, "No channel layout specified. The encoder "
166 "will use Vorbis channel layout for "
167 "%d channels.\n", avctx->channels);
168 }
169 }
170
171 if ((ret = vorbis_encode_setup_init(vi)))
172 goto error;
173
174 return 0;
175 error:
176 return vorbis_error_to_averror(ret);
177 }
178
179 /* How many bytes are needed for a buffer of length 'l' */
xiph_len(int l)180 static int xiph_len(int l)
181 {
182 return 1 + l / 255 + l;
183 }
184
libvorbis_encode_close(AVCodecContext * avctx)185 static av_cold int libvorbis_encode_close(AVCodecContext *avctx)
186 {
187 LibvorbisEncContext *s = avctx->priv_data;
188
189 /* notify vorbisenc this is EOF */
190 if (s->dsp_initialized)
191 vorbis_analysis_wrote(&s->vd, 0);
192
193 vorbis_block_clear(&s->vb);
194 vorbis_dsp_clear(&s->vd);
195 vorbis_info_clear(&s->vi);
196
197 av_fifo_freep(&s->pkt_fifo);
198 ff_af_queue_close(&s->afq);
199 av_freep(&avctx->extradata);
200
201 av_vorbis_parse_free(&s->vp);
202
203 return 0;
204 }
205
libvorbis_encode_init(AVCodecContext * avctx)206 static av_cold int libvorbis_encode_init(AVCodecContext *avctx)
207 {
208 LibvorbisEncContext *s = avctx->priv_data;
209 ogg_packet header, header_comm, header_code;
210 uint8_t *p;
211 unsigned int offset;
212 int ret;
213
214 vorbis_info_init(&s->vi);
215 if ((ret = libvorbis_setup(&s->vi, avctx))) {
216 av_log(avctx, AV_LOG_ERROR, "encoder setup failed\n");
217 goto error;
218 }
219 if ((ret = vorbis_analysis_init(&s->vd, &s->vi))) {
220 av_log(avctx, AV_LOG_ERROR, "analysis init failed\n");
221 ret = vorbis_error_to_averror(ret);
222 goto error;
223 }
224 s->dsp_initialized = 1;
225 if ((ret = vorbis_block_init(&s->vd, &s->vb))) {
226 av_log(avctx, AV_LOG_ERROR, "dsp init failed\n");
227 ret = vorbis_error_to_averror(ret);
228 goto error;
229 }
230
231 vorbis_comment_init(&s->vc);
232 if (!(avctx->flags & AV_CODEC_FLAG_BITEXACT))
233 vorbis_comment_add_tag(&s->vc, "encoder", LIBAVCODEC_IDENT);
234
235 if ((ret = vorbis_analysis_headerout(&s->vd, &s->vc, &header, &header_comm,
236 &header_code))) {
237 ret = vorbis_error_to_averror(ret);
238 goto error;
239 }
240
241 avctx->extradata_size = 1 + xiph_len(header.bytes) +
242 xiph_len(header_comm.bytes) +
243 header_code.bytes;
244 p = avctx->extradata = av_malloc(avctx->extradata_size +
245 AV_INPUT_BUFFER_PADDING_SIZE);
246 if (!p) {
247 ret = AVERROR(ENOMEM);
248 goto error;
249 }
250 p[0] = 2;
251 offset = 1;
252 offset += av_xiphlacing(&p[offset], header.bytes);
253 offset += av_xiphlacing(&p[offset], header_comm.bytes);
254 memcpy(&p[offset], header.packet, header.bytes);
255 offset += header.bytes;
256 memcpy(&p[offset], header_comm.packet, header_comm.bytes);
257 offset += header_comm.bytes;
258 memcpy(&p[offset], header_code.packet, header_code.bytes);
259 offset += header_code.bytes;
260 av_assert0(offset == avctx->extradata_size);
261
262 s->vp = av_vorbis_parse_init(avctx->extradata, avctx->extradata_size);
263 if (!s->vp) {
264 av_log(avctx, AV_LOG_ERROR, "invalid extradata\n");
265 return ret;
266 }
267
268 vorbis_comment_clear(&s->vc);
269
270 avctx->frame_size = LIBVORBIS_FRAME_SIZE;
271 ff_af_queue_init(avctx, &s->afq);
272
273 s->pkt_fifo = av_fifo_alloc(BUFFER_SIZE);
274 if (!s->pkt_fifo) {
275 ret = AVERROR(ENOMEM);
276 goto error;
277 }
278
279 return 0;
280 error:
281 libvorbis_encode_close(avctx);
282 return ret;
283 }
284
libvorbis_encode_frame(AVCodecContext * avctx,AVPacket * avpkt,const AVFrame * frame,int * got_packet_ptr)285 static int libvorbis_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
286 const AVFrame *frame, int *got_packet_ptr)
287 {
288 LibvorbisEncContext *s = avctx->priv_data;
289 ogg_packet op;
290 int ret, duration;
291
292 /* send samples to libvorbis */
293 if (frame) {
294 const int samples = frame->nb_samples;
295 float **buffer;
296 int c, channels = s->vi.channels;
297
298 buffer = vorbis_analysis_buffer(&s->vd, samples);
299 for (c = 0; c < channels; c++) {
300 int co = (channels > 8) ? c :
301 vorbis_encoding_channel_layout_offsets[channels - 1][c];
302 memcpy(buffer[c], frame->extended_data[co],
303 samples * sizeof(*buffer[c]));
304 }
305 if ((ret = vorbis_analysis_wrote(&s->vd, samples)) < 0) {
306 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
307 return vorbis_error_to_averror(ret);
308 }
309 if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
310 return ret;
311 } else {
312 if (!s->eof && s->afq.frame_alloc)
313 if ((ret = vorbis_analysis_wrote(&s->vd, 0)) < 0) {
314 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
315 return vorbis_error_to_averror(ret);
316 }
317 s->eof = 1;
318 }
319
320 /* retrieve available packets from libvorbis */
321 while ((ret = vorbis_analysis_blockout(&s->vd, &s->vb)) == 1) {
322 if ((ret = vorbis_analysis(&s->vb, NULL)) < 0)
323 break;
324 if ((ret = vorbis_bitrate_addblock(&s->vb)) < 0)
325 break;
326
327 /* add any available packets to the output packet buffer */
328 while ((ret = vorbis_bitrate_flushpacket(&s->vd, &op)) == 1) {
329 if (av_fifo_space(s->pkt_fifo) < sizeof(ogg_packet) + op.bytes) {
330 av_log(avctx, AV_LOG_ERROR, "packet buffer is too small\n");
331 return AVERROR_BUG;
332 }
333 av_fifo_generic_write(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
334 av_fifo_generic_write(s->pkt_fifo, op.packet, op.bytes, NULL);
335 }
336 if (ret < 0) {
337 av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
338 break;
339 }
340 }
341 if (ret < 0) {
342 av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
343 return vorbis_error_to_averror(ret);
344 }
345
346 /* check for available packets */
347 if (av_fifo_size(s->pkt_fifo) < sizeof(ogg_packet))
348 return 0;
349
350 av_fifo_generic_read(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
351
352 if ((ret = ff_alloc_packet2(avctx, avpkt, op.bytes, 0)) < 0)
353 return ret;
354 av_fifo_generic_read(s->pkt_fifo, avpkt->data, op.bytes, NULL);
355
356 avpkt->pts = ff_samples_to_time_base(avctx, op.granulepos);
357
358 duration = av_vorbis_parse_frame(s->vp, avpkt->data, avpkt->size);
359 if (duration > 0) {
360 /* we do not know encoder delay until we get the first packet from
361 * libvorbis, so we have to update the AudioFrameQueue counts */
362 if (!avctx->initial_padding && s->afq.frames) {
363 avctx->initial_padding = duration;
364 av_assert0(!s->afq.remaining_delay);
365 s->afq.frames->duration += duration;
366 if (s->afq.frames->pts != AV_NOPTS_VALUE)
367 s->afq.frames->pts -= duration;
368 s->afq.remaining_samples += duration;
369 }
370 ff_af_queue_remove(&s->afq, duration, &avpkt->pts, &avpkt->duration);
371 }
372
373 *got_packet_ptr = 1;
374 return 0;
375 }
376
377 AVCodec ff_libvorbis_encoder = {
378 .name = "libvorbis",
379 .long_name = NULL_IF_CONFIG_SMALL("libvorbis"),
380 .type = AVMEDIA_TYPE_AUDIO,
381 .id = AV_CODEC_ID_VORBIS,
382 .priv_data_size = sizeof(LibvorbisEncContext),
383 .init = libvorbis_encode_init,
384 .encode2 = libvorbis_encode_frame,
385 .close = libvorbis_encode_close,
386 .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_SMALL_LAST_FRAME,
387 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
388 AV_SAMPLE_FMT_NONE },
389 .priv_class = &vorbis_class,
390 .defaults = defaults,
391 .wrapper_name = "libvorbis",
392 };
393