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
21 #include <vorbis/vorbisenc.h>
23 #include "libavutil/avassert.h"
24 #include "libavutil/fifo.h"
25 #include "libavutil/opt.h"
27 #include "audio_frame_queue.h"
30 #include "vorbis_parser.h"
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
38 #define OGGVORBIS_FRAME_SIZE 64
40 #define BUFFER_SIZE (1024 * 64)
42 typedef struct OggVorbisEncContext {
43 AVClass *av_class; /**< class for AVOptions */
45 vorbis_info vi; /**< vorbis_info used during init */
46 vorbis_dsp_state vd; /**< DSP state used for analysis */
47 vorbis_block vb; /**< vorbis_block used for analysis */
48 AVFifoBuffer *pkt_fifo; /**< output packet buffer */
49 int eof; /**< end-of-file flag */
50 int dsp_initialized; /**< vd has been initialized */
51 vorbis_comment vc; /**< VorbisComment info */
52 double iblock; /**< impulse block bias option */
53 VorbisParseContext vp; /**< parse context to get durations */
54 AudioFrameQueue afq; /**< frame queue for timestamps */
55 } OggVorbisEncContext;
57 static const AVOption options[] = {
58 { "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 },
62 static const AVCodecDefault defaults[] = {
67 static const AVClass class = {
68 .class_name = "libvorbis",
69 .item_name = av_default_item_name,
71 .version = LIBAVUTIL_VERSION_INT,
74 static int vorbis_error_to_averror(int ov_err)
77 case OV_EFAULT: return AVERROR_BUG;
78 case OV_EINVAL: return AVERROR(EINVAL);
79 case OV_EIMPL: return AVERROR(EINVAL);
80 default: return AVERROR_UNKNOWN;
84 static av_cold int oggvorbis_init_encoder(vorbis_info *vi,
85 AVCodecContext *avctx)
87 OggVorbisEncContext *s = avctx->priv_data;
91 if (avctx->flags & CODEC_FLAG_QSCALE || !avctx->bit_rate) {
93 * NOTE: we use the oggenc range of -1 to 10 for global_quality for
94 * user convenience, but libvorbis uses -0.1 to 1.0.
96 float q = avctx->global_quality / (float)FF_QP2LAMBDA;
97 /* default to 3 if the user did not set quality or bitrate */
98 if (!(avctx->flags & CODEC_FLAG_QSCALE))
100 if ((ret = vorbis_encode_setup_vbr(vi, avctx->channels,
105 int minrate = avctx->rc_min_rate > 0 ? avctx->rc_min_rate : -1;
106 int maxrate = avctx->rc_max_rate > 0 ? avctx->rc_max_rate : -1;
108 /* average bitrate */
109 if ((ret = vorbis_encode_setup_managed(vi, avctx->channels,
110 avctx->sample_rate, maxrate,
111 avctx->bit_rate, minrate)))
114 /* variable bitrate by estimate, disable slow rate management */
115 if (minrate == -1 && maxrate == -1)
116 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL)))
117 goto error; /* should not happen */
120 /* cutoff frequency */
121 if (avctx->cutoff > 0) {
122 cfreq = avctx->cutoff / 1000.0;
123 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq)))
124 goto error; /* should not happen */
127 /* impulse block bias */
129 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &s->iblock)))
133 if (avctx->channels == 3 &&
134 avctx->channel_layout != (AV_CH_LAYOUT_STEREO|AV_CH_FRONT_CENTER) ||
135 avctx->channels == 4 &&
136 avctx->channel_layout != AV_CH_LAYOUT_2_2 &&
137 avctx->channel_layout != AV_CH_LAYOUT_QUAD ||
138 avctx->channels == 5 &&
139 avctx->channel_layout != AV_CH_LAYOUT_5POINT0 &&
140 avctx->channel_layout != AV_CH_LAYOUT_5POINT0_BACK ||
141 avctx->channels == 6 &&
142 avctx->channel_layout != AV_CH_LAYOUT_5POINT1 &&
143 avctx->channel_layout != AV_CH_LAYOUT_5POINT1_BACK ||
144 avctx->channels == 7 &&
145 avctx->channel_layout != (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_CENTER) ||
146 avctx->channels == 8 &&
147 avctx->channel_layout != AV_CH_LAYOUT_7POINT1) {
148 if (avctx->channel_layout) {
150 av_get_channel_layout_string(name, sizeof(name), avctx->channels,
151 avctx->channel_layout);
152 av_log(avctx, AV_LOG_ERROR, "%s not supported by Vorbis: "
153 "output stream will have incorrect "
154 "channel layout.\n", name);
156 av_log(avctx, AV_LOG_WARNING, "No channel layout specified. The encoder "
157 "will use Vorbis channel layout for "
158 "%d channels.\n", avctx->channels);
162 if ((ret = vorbis_encode_setup_init(vi)))
167 return vorbis_error_to_averror(ret);
170 /* How many bytes are needed for a buffer of length 'l' */
171 static int xiph_len(int l)
173 return 1 + l / 255 + l;
176 static av_cold int oggvorbis_encode_close(AVCodecContext *avctx)
178 OggVorbisEncContext *s = avctx->priv_data;
180 /* notify vorbisenc this is EOF */
181 if (s->dsp_initialized)
182 vorbis_analysis_wrote(&s->vd, 0);
184 vorbis_block_clear(&s->vb);
185 vorbis_dsp_clear(&s->vd);
186 vorbis_info_clear(&s->vi);
188 av_fifo_free(s->pkt_fifo);
189 ff_af_queue_close(&s->afq);
190 #if FF_API_OLD_ENCODE_AUDIO
191 av_freep(&avctx->coded_frame);
193 av_freep(&avctx->extradata);
198 static av_cold int oggvorbis_encode_init(AVCodecContext *avctx)
200 OggVorbisEncContext *s = avctx->priv_data;
201 ogg_packet header, header_comm, header_code;
206 vorbis_info_init(&s->vi);
207 if ((ret = oggvorbis_init_encoder(&s->vi, avctx))) {
208 av_log(avctx, AV_LOG_ERROR, "encoder setup failed\n");
211 if ((ret = vorbis_analysis_init(&s->vd, &s->vi))) {
212 av_log(avctx, AV_LOG_ERROR, "analysis init failed\n");
213 ret = vorbis_error_to_averror(ret);
216 s->dsp_initialized = 1;
217 if ((ret = vorbis_block_init(&s->vd, &s->vb))) {
218 av_log(avctx, AV_LOG_ERROR, "dsp init failed\n");
219 ret = vorbis_error_to_averror(ret);
223 vorbis_comment_init(&s->vc);
224 if (!(avctx->flags & CODEC_FLAG_BITEXACT))
225 vorbis_comment_add_tag(&s->vc, "encoder", LIBAVCODEC_IDENT);
227 if ((ret = vorbis_analysis_headerout(&s->vd, &s->vc, &header, &header_comm,
229 ret = vorbis_error_to_averror(ret);
233 avctx->extradata_size = 1 + xiph_len(header.bytes) +
234 xiph_len(header_comm.bytes) +
236 p = avctx->extradata = av_malloc(avctx->extradata_size +
237 FF_INPUT_BUFFER_PADDING_SIZE);
239 ret = AVERROR(ENOMEM);
244 offset += av_xiphlacing(&p[offset], header.bytes);
245 offset += av_xiphlacing(&p[offset], header_comm.bytes);
246 memcpy(&p[offset], header.packet, header.bytes);
247 offset += header.bytes;
248 memcpy(&p[offset], header_comm.packet, header_comm.bytes);
249 offset += header_comm.bytes;
250 memcpy(&p[offset], header_code.packet, header_code.bytes);
251 offset += header_code.bytes;
252 av_assert0(offset == avctx->extradata_size);
254 if ((ret = avpriv_vorbis_parse_extradata(avctx, &s->vp)) < 0) {
255 av_log(avctx, AV_LOG_ERROR, "invalid extradata\n");
259 vorbis_comment_clear(&s->vc);
261 avctx->frame_size = OGGVORBIS_FRAME_SIZE;
262 ff_af_queue_init(avctx, &s->afq);
264 s->pkt_fifo = av_fifo_alloc(BUFFER_SIZE);
266 ret = AVERROR(ENOMEM);
270 #if FF_API_OLD_ENCODE_AUDIO
271 avctx->coded_frame = avcodec_alloc_frame();
272 if (!avctx->coded_frame) {
273 ret = AVERROR(ENOMEM);
280 oggvorbis_encode_close(avctx);
284 static int oggvorbis_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
285 const AVFrame *frame, int *got_packet_ptr)
287 OggVorbisEncContext *s = avctx->priv_data;
291 /* send samples to libvorbis */
293 const int samples = frame->nb_samples;
295 int c, channels = s->vi.channels;
297 buffer = vorbis_analysis_buffer(&s->vd, samples);
298 for (c = 0; c < channels; c++) {
299 int co = (channels > 8) ? c :
300 ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
301 memcpy(buffer[c], frame->extended_data[co],
302 samples * sizeof(*buffer[c]));
304 if ((ret = vorbis_analysis_wrote(&s->vd, samples)) < 0) {
305 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
306 return vorbis_error_to_averror(ret);
308 if ((ret = ff_af_queue_add(&s->afq, frame) < 0))
312 if ((ret = vorbis_analysis_wrote(&s->vd, 0)) < 0) {
313 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
314 return vorbis_error_to_averror(ret);
319 /* retrieve available packets from libvorbis */
320 while ((ret = vorbis_analysis_blockout(&s->vd, &s->vb)) == 1) {
321 if ((ret = vorbis_analysis(&s->vb, NULL)) < 0)
323 if ((ret = vorbis_bitrate_addblock(&s->vb)) < 0)
326 /* add any available packets to the output packet buffer */
327 while ((ret = vorbis_bitrate_flushpacket(&s->vd, &op)) == 1) {
328 if (av_fifo_space(s->pkt_fifo) < sizeof(ogg_packet) + op.bytes) {
329 av_log(avctx, AV_LOG_ERROR, "packet buffer is too small\n");
332 av_fifo_generic_write(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
333 av_fifo_generic_write(s->pkt_fifo, op.packet, op.bytes, NULL);
336 av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
341 av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
342 return vorbis_error_to_averror(ret);
345 /* check for available packets */
346 if (av_fifo_size(s->pkt_fifo) < sizeof(ogg_packet))
349 av_fifo_generic_read(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
351 if ((ret = ff_alloc_packet2(avctx, avpkt, op.bytes)))
353 av_fifo_generic_read(s->pkt_fifo, avpkt->data, op.bytes, NULL);
355 avpkt->pts = ff_samples_to_time_base(avctx, op.granulepos);
357 duration = avpriv_vorbis_parse_frame(&s->vp, avpkt->data, avpkt->size);
359 /* we do not know encoder delay until we get the first packet from
360 * libvorbis, so we have to update the AudioFrameQueue counts */
362 avctx->delay = duration;
363 av_assert0(!s->afq.remaining_delay);
364 s->afq.frames->duration += duration;
365 s->afq.frames->pts -= duration;
366 s->afq.remaining_samples += duration;
368 ff_af_queue_remove(&s->afq, duration, &avpkt->pts, &avpkt->duration);
375 AVCodec ff_libvorbis_encoder = {
377 .type = AVMEDIA_TYPE_AUDIO,
378 .id = AV_CODEC_ID_VORBIS,
379 .priv_data_size = sizeof(OggVorbisEncContext),
380 .init = oggvorbis_encode_init,
381 .encode2 = oggvorbis_encode_frame,
382 .close = oggvorbis_encode_close,
383 .capabilities = CODEC_CAP_DELAY,
384 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
385 AV_SAMPLE_FMT_NONE },
386 .long_name = NULL_IF_CONFIG_SMALL("libvorbis"),
387 .priv_class = &class,
388 .defaults = defaults,