2 * copyright (c) 2002 Mark Hills <mark@pogo.org.uk>
4 * This file is part of FFmpeg.
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.
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.
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
23 * Vorbis encoding support via libvorbisenc.
24 * @author Mark Hills <mark@pogo.org.uk>
27 #include <vorbis/vorbisenc.h>
29 #include "libavutil/fifo.h"
30 #include "libavutil/opt.h"
32 #include "audio_frame_queue.h"
33 #include "bytestream.h"
36 #include "vorbis_parser.h"
41 /* Number of samples the user should send in each call.
42 * This value is used because it is the LCD of all possible frame sizes, so
43 * an output packet will always start at the same point as one of the input
46 #define OGGVORBIS_FRAME_SIZE 64
48 #define BUFFER_SIZE (1024 * 64)
50 typedef struct OggVorbisContext {
51 AVClass *av_class; /**< class for AVOptions */
52 vorbis_info vi; /**< vorbis_info used during init */
53 vorbis_dsp_state vd; /**< DSP state used for analysis */
54 vorbis_block vb; /**< vorbis_block used for analysis */
55 AVFifoBuffer *pkt_fifo; /**< output packet buffer */
56 int eof; /**< end-of-file flag */
57 int dsp_initialized; /**< vd has been initialized */
58 vorbis_comment vc; /**< VorbisComment info */
59 ogg_packet op; /**< ogg packet */
60 double iblock; /**< impulse block bias option */
61 VorbisParseContext vp; /**< parse context to get durations */
62 AudioFrameQueue afq; /**< frame queue for timestamps */
65 static const AVOption options[] = {
66 { "iblock", "Sets the impulse block bias", offsetof(OggVorbisContext, iblock), AV_OPT_TYPE_DOUBLE, { .dbl = 0 }, -15, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM },
70 static const AVCodecDefault defaults[] = {
75 static const AVClass class = { "libvorbis", av_default_item_name, options, LIBAVUTIL_VERSION_INT };
78 static int vorbis_error_to_averror(int ov_err)
81 case OV_EFAULT: return AVERROR_BUG;
82 case OV_EINVAL: return AVERROR(EINVAL);
83 case OV_EIMPL: return AVERROR(EINVAL);
84 default: return AVERROR_UNKNOWN;
88 static av_cold int oggvorbis_init_encoder(vorbis_info *vi,
89 AVCodecContext *avctx)
91 OggVorbisContext *s = avctx->priv_data;
95 if (avctx->flags & CODEC_FLAG_QSCALE || !avctx->bit_rate) {
97 * NOTE: we use the oggenc range of -1 to 10 for global_quality for
98 * user convenience, but libvorbis uses -0.1 to 1.0.
100 float q = avctx->global_quality / (float)FF_QP2LAMBDA;
101 /* default to 3 if the user did not set quality or bitrate */
102 if (!(avctx->flags & CODEC_FLAG_QSCALE))
104 if ((ret = vorbis_encode_setup_vbr(vi, avctx->channels,
109 int minrate = avctx->rc_min_rate > 0 ? avctx->rc_min_rate : -1;
110 int maxrate = avctx->rc_max_rate > 0 ? avctx->rc_max_rate : -1;
112 /* average bitrate */
113 if ((ret = vorbis_encode_setup_managed(vi, avctx->channels,
114 avctx->sample_rate, maxrate,
115 avctx->bit_rate, minrate)))
118 /* variable bitrate by estimate, disable slow rate management */
119 if (minrate == -1 && maxrate == -1)
120 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL)))
121 goto error; /* should not happen */
124 /* cutoff frequency */
125 if (avctx->cutoff > 0) {
126 cfreq = avctx->cutoff / 1000.0;
127 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq)))
128 goto error; /* should not happen */
131 /* impulse block bias */
133 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &s->iblock)))
137 if (avctx->channels == 3 &&
138 avctx->channel_layout != (AV_CH_LAYOUT_STEREO|AV_CH_FRONT_CENTER) ||
139 avctx->channels == 4 &&
140 avctx->channel_layout != AV_CH_LAYOUT_2_2 &&
141 avctx->channel_layout != AV_CH_LAYOUT_QUAD ||
142 avctx->channels == 5 &&
143 avctx->channel_layout != AV_CH_LAYOUT_5POINT0 &&
144 avctx->channel_layout != AV_CH_LAYOUT_5POINT0_BACK ||
145 avctx->channels == 6 &&
146 avctx->channel_layout != AV_CH_LAYOUT_5POINT1 &&
147 avctx->channel_layout != AV_CH_LAYOUT_5POINT1_BACK ||
148 avctx->channels == 7 &&
149 avctx->channel_layout != (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_CENTER) ||
150 avctx->channels == 8 &&
151 avctx->channel_layout != AV_CH_LAYOUT_7POINT1) {
152 if (avctx->channel_layout) {
154 av_get_channel_layout_string(name, sizeof(name), avctx->channels,
155 avctx->channel_layout);
156 av_log(avctx, AV_LOG_ERROR, "%s not supported by Vorbis: "
157 "output stream will have incorrect "
158 "channel layout.\n", name);
160 av_log(avctx, AV_LOG_WARNING, "No channel layout specified. The encoder "
161 "will use Vorbis channel layout for "
162 "%d channels.\n", avctx->channels);
166 if ((ret = vorbis_encode_setup_init(vi)))
171 return vorbis_error_to_averror(ret);
174 /* How many bytes are needed for a buffer of length 'l' */
175 static int xiph_len(int l)
177 return 1 + l / 255 + l;
180 static av_cold int oggvorbis_encode_close(AVCodecContext *avctx)
182 OggVorbisContext *s = avctx->priv_data;
184 /* notify vorbisenc this is EOF */
185 if (s->dsp_initialized)
186 vorbis_analysis_wrote(&s->vd, 0);
188 vorbis_block_clear(&s->vb);
189 vorbis_dsp_clear(&s->vd);
190 vorbis_info_clear(&s->vi);
192 av_fifo_free(s->pkt_fifo);
193 ff_af_queue_close(&s->afq);
194 #if FF_API_OLD_ENCODE_AUDIO
195 av_freep(&avctx->coded_frame);
197 av_freep(&avctx->extradata);
202 static av_cold int oggvorbis_encode_init(AVCodecContext *avctx)
204 OggVorbisContext *s = avctx->priv_data;
205 ogg_packet header, header_comm, header_code;
210 vorbis_info_init(&s->vi);
211 if ((ret = oggvorbis_init_encoder(&s->vi, avctx))) {
212 av_log(avctx, AV_LOG_ERROR, "encoder setup failed\n");
215 if ((ret = vorbis_analysis_init(&s->vd, &s->vi))) {
216 av_log(avctx, AV_LOG_ERROR, "analysis init failed\n");
217 ret = vorbis_error_to_averror(ret);
220 s->dsp_initialized = 1;
221 if ((ret = vorbis_block_init(&s->vd, &s->vb))) {
222 av_log(avctx, AV_LOG_ERROR, "dsp init failed\n");
223 ret = vorbis_error_to_averror(ret);
227 vorbis_comment_init(&s->vc);
228 vorbis_comment_add_tag(&s->vc, "encoder", LIBAVCODEC_IDENT);
230 if ((ret = vorbis_analysis_headerout(&s->vd, &s->vc, &header, &header_comm,
232 ret = vorbis_error_to_averror(ret);
236 avctx->extradata_size = 1 + xiph_len(header.bytes) +
237 xiph_len(header_comm.bytes) +
239 p = avctx->extradata = av_malloc(avctx->extradata_size +
240 FF_INPUT_BUFFER_PADDING_SIZE);
242 ret = AVERROR(ENOMEM);
247 offset += av_xiphlacing(&p[offset], header.bytes);
248 offset += av_xiphlacing(&p[offset], header_comm.bytes);
249 memcpy(&p[offset], header.packet, header.bytes);
250 offset += header.bytes;
251 memcpy(&p[offset], header_comm.packet, header_comm.bytes);
252 offset += header_comm.bytes;
253 memcpy(&p[offset], header_code.packet, header_code.bytes);
254 offset += header_code.bytes;
255 assert(offset == avctx->extradata_size);
257 if ((ret = avpriv_vorbis_parse_extradata(avctx, &s->vp)) < 0) {
258 av_log(avctx, AV_LOG_ERROR, "invalid extradata\n");
262 vorbis_comment_clear(&s->vc);
264 avctx->frame_size = OGGVORBIS_FRAME_SIZE;
265 ff_af_queue_init(avctx, &s->afq);
267 s->pkt_fifo = av_fifo_alloc(BUFFER_SIZE);
269 ret = AVERROR(ENOMEM);
273 #if FF_API_OLD_ENCODE_AUDIO
274 avctx->coded_frame = avcodec_alloc_frame();
275 if (!avctx->coded_frame) {
276 ret = AVERROR(ENOMEM);
283 oggvorbis_encode_close(avctx);
287 static int oggvorbis_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
288 const AVFrame *frame, int *got_packet_ptr)
290 OggVorbisContext *s = avctx->priv_data;
294 /* send samples to libvorbis */
296 const float *audio = (const float *)frame->data[0];
297 const int samples = frame->nb_samples;
299 int c, channels = s->vi.channels;
301 buffer = vorbis_analysis_buffer(&s->vd, samples);
302 for (c = 0; c < channels; c++) {
304 int co = (channels > 8) ? c :
305 ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
306 for (i = 0; i < samples; i++)
307 buffer[c][i] = audio[i * channels + co];
309 if ((ret = vorbis_analysis_wrote(&s->vd, samples)) < 0) {
310 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
311 return vorbis_error_to_averror(ret);
313 if ((ret = ff_af_queue_add(&s->afq, frame) < 0))
317 if ((ret = vorbis_analysis_wrote(&s->vd, 0)) < 0) {
318 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
319 return vorbis_error_to_averror(ret);
324 /* retrieve available packets from libvorbis */
325 while ((ret = vorbis_analysis_blockout(&s->vd, &s->vb)) == 1) {
326 if ((ret = vorbis_analysis(&s->vb, NULL)) < 0)
328 if ((ret = vorbis_bitrate_addblock(&s->vb)) < 0)
331 /* add any available packets to the output packet buffer */
332 while ((ret = vorbis_bitrate_flushpacket(&s->vd, &op)) == 1) {
333 if (av_fifo_space(s->pkt_fifo) < sizeof(ogg_packet) + op.bytes) {
334 av_log(avctx, AV_LOG_ERROR, "packet buffer is too small\n");
337 av_fifo_generic_write(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
338 av_fifo_generic_write(s->pkt_fifo, op.packet, op.bytes, NULL);
341 av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
346 av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
347 return vorbis_error_to_averror(ret);
350 /* check for available packets */
351 if (av_fifo_size(s->pkt_fifo) < sizeof(ogg_packet))
354 av_fifo_generic_read(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
356 if ((ret = ff_alloc_packet2(avctx, avpkt, op.bytes)))
358 av_fifo_generic_read(s->pkt_fifo, avpkt->data, op.bytes, NULL);
360 avpkt->pts = ff_samples_to_time_base(avctx, op.granulepos);
362 duration = avpriv_vorbis_parse_frame(&s->vp, avpkt->data, avpkt->size);
364 /* we do not know encoder delay until we get the first packet from
365 * libvorbis, so we have to update the AudioFrameQueue counts */
367 avctx->delay = duration;
368 s->afq.remaining_delay += duration;
369 s->afq.remaining_samples += duration;
371 ff_af_queue_remove(&s->afq, duration, &avpkt->pts, &avpkt->duration);
378 AVCodec ff_libvorbis_encoder = {
380 .type = AVMEDIA_TYPE_AUDIO,
381 .id = CODEC_ID_VORBIS,
382 .priv_data_size = sizeof(OggVorbisContext),
383 .init = oggvorbis_encode_init,
384 .encode2 = oggvorbis_encode_frame,
385 .close = oggvorbis_encode_close,
386 .capabilities = CODEC_CAP_DELAY,
387 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLT,
388 AV_SAMPLE_FMT_NONE },
389 .long_name = NULL_IF_CONFIG_SMALL("libvorbis Vorbis"),
390 .priv_class = &class,
391 .defaults = defaults,