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 */
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 */
54 } OggVorbisEncContext;
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 },
61 static const AVCodecDefault defaults[] = {
66 static const AVClass vorbis_class = {
67 .class_name = "libvorbis",
68 .item_name = av_default_item_name,
70 .version = LIBAVUTIL_VERSION_INT,
73 static int vorbis_error_to_averror(int 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;
83 static av_cold int oggvorbis_encode_init_internal(vorbis_info *vi, AVCodecContext *avctx)
85 OggVorbisEncContext *s = avctx->priv_data;
89 if (avctx->flags & CODEC_FLAG_QSCALE || !avctx->bit_rate) {
91 * NOTE: we use the oggenc range of -1 to 10 for global_quality for
92 * user convenience, but libvorbis uses -0.1 to 1.0.
94 float q = avctx->global_quality / (float)FF_QP2LAMBDA;
95 /* default to 3 if the user did not set quality or bitrate */
96 if (!(avctx->flags & CODEC_FLAG_QSCALE))
98 if ((ret = vorbis_encode_setup_vbr(vi, avctx->channels,
103 int minrate = avctx->rc_min_rate > 0 ? avctx->rc_min_rate : -1;
104 int maxrate = avctx->rc_max_rate > 0 ? avctx->rc_max_rate : -1;
106 /* average bitrate */
107 if ((ret = vorbis_encode_setup_managed(vi, avctx->channels,
108 avctx->sample_rate, maxrate,
109 avctx->bit_rate, minrate)))
112 /* variable bitrate by estimate, disable slow rate management */
113 if (minrate == -1 && maxrate == -1)
114 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL)))
115 goto error; /* should not happen */
118 /* cutoff frequency */
119 if (avctx->cutoff > 0) {
120 cfreq = avctx->cutoff / 1000.0;
121 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq)))
122 goto error; /* should not happen */
125 /* impulse block bias */
127 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &s->iblock)))
131 if (avctx->channels == 3 &&
132 avctx->channel_layout != (AV_CH_LAYOUT_STEREO|AV_CH_FRONT_CENTER) ||
133 avctx->channels == 4 &&
134 avctx->channel_layout != AV_CH_LAYOUT_2_2 &&
135 avctx->channel_layout != AV_CH_LAYOUT_QUAD ||
136 avctx->channels == 5 &&
137 avctx->channel_layout != AV_CH_LAYOUT_5POINT0 &&
138 avctx->channel_layout != AV_CH_LAYOUT_5POINT0_BACK ||
139 avctx->channels == 6 &&
140 avctx->channel_layout != AV_CH_LAYOUT_5POINT1 &&
141 avctx->channel_layout != AV_CH_LAYOUT_5POINT1_BACK ||
142 avctx->channels == 7 &&
143 avctx->channel_layout != (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_CENTER) ||
144 avctx->channels == 8 &&
145 avctx->channel_layout != AV_CH_LAYOUT_7POINT1) {
146 if (avctx->channel_layout) {
148 av_get_channel_layout_string(name, sizeof(name), avctx->channels,
149 avctx->channel_layout);
150 av_log(avctx, AV_LOG_ERROR, "%s not supported by Vorbis: "
151 "output stream will have incorrect "
152 "channel layout.\n", name);
154 av_log(avctx, AV_LOG_WARNING, "No channel layout specified. The encoder "
155 "will use Vorbis channel layout for "
156 "%d channels.\n", avctx->channels);
160 if ((ret = vorbis_encode_setup_init(vi)))
165 return vorbis_error_to_averror(ret);
168 /* How many bytes are needed for a buffer of length 'l' */
169 static int xiph_len(int l)
171 return 1 + l / 255 + l;
174 static av_cold int oggvorbis_encode_close(AVCodecContext *avctx)
176 OggVorbisEncContext *s = avctx->priv_data;
178 /* notify vorbisenc this is EOF */
179 if (s->dsp_initialized)
180 vorbis_analysis_wrote(&s->vd, 0);
182 vorbis_block_clear(&s->vb);
183 vorbis_dsp_clear(&s->vd);
184 vorbis_info_clear(&s->vi);
186 av_fifo_free(s->pkt_fifo);
187 ff_af_queue_close(&s->afq);
188 av_freep(&avctx->extradata);
193 static av_cold int oggvorbis_encode_init(AVCodecContext *avctx)
195 OggVorbisEncContext *s = avctx->priv_data;
196 ogg_packet header, header_comm, header_code;
201 vorbis_info_init(&s->vi);
202 if ((ret = oggvorbis_encode_init_internal(&s->vi, avctx))) {
203 av_log(avctx, AV_LOG_ERROR, "encoder setup failed\n");
206 if ((ret = vorbis_analysis_init(&s->vd, &s->vi))) {
207 av_log(avctx, AV_LOG_ERROR, "analysis init failed\n");
208 ret = vorbis_error_to_averror(ret);
211 s->dsp_initialized = 1;
212 if ((ret = vorbis_block_init(&s->vd, &s->vb))) {
213 av_log(avctx, AV_LOG_ERROR, "dsp init failed\n");
214 ret = vorbis_error_to_averror(ret);
218 vorbis_comment_init(&s->vc);
219 if (!(avctx->flags & CODEC_FLAG_BITEXACT))
220 vorbis_comment_add_tag(&s->vc, "encoder", LIBAVCODEC_IDENT);
222 if ((ret = vorbis_analysis_headerout(&s->vd, &s->vc, &header, &header_comm,
224 ret = vorbis_error_to_averror(ret);
228 avctx->extradata_size = 1 + xiph_len(header.bytes) +
229 xiph_len(header_comm.bytes) +
231 p = avctx->extradata = av_malloc(avctx->extradata_size +
232 FF_INPUT_BUFFER_PADDING_SIZE);
234 ret = AVERROR(ENOMEM);
239 offset += av_xiphlacing(&p[offset], header.bytes);
240 offset += av_xiphlacing(&p[offset], header_comm.bytes);
241 memcpy(&p[offset], header.packet, header.bytes);
242 offset += header.bytes;
243 memcpy(&p[offset], header_comm.packet, header_comm.bytes);
244 offset += header_comm.bytes;
245 memcpy(&p[offset], header_code.packet, header_code.bytes);
246 offset += header_code.bytes;
247 av_assert0(offset == avctx->extradata_size);
249 if ((ret = avpriv_vorbis_parse_extradata(avctx, &s->vp)) < 0) {
250 av_log(avctx, AV_LOG_ERROR, "invalid extradata\n");
254 vorbis_comment_clear(&s->vc);
256 avctx->frame_size = OGGVORBIS_FRAME_SIZE;
257 ff_af_queue_init(avctx, &s->afq);
259 s->pkt_fifo = av_fifo_alloc(BUFFER_SIZE);
261 ret = AVERROR(ENOMEM);
267 oggvorbis_encode_close(avctx);
271 static int oggvorbis_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
272 const AVFrame *frame, int *got_packet_ptr)
274 OggVorbisEncContext *s = avctx->priv_data;
278 /* send samples to libvorbis */
280 const int samples = frame->nb_samples;
282 int c, channels = s->vi.channels;
284 buffer = vorbis_analysis_buffer(&s->vd, samples);
285 for (c = 0; c < channels; c++) {
286 int co = (channels > 8) ? c :
287 ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
288 memcpy(buffer[c], frame->extended_data[co],
289 samples * sizeof(*buffer[c]));
291 if ((ret = vorbis_analysis_wrote(&s->vd, samples)) < 0) {
292 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
293 return vorbis_error_to_averror(ret);
295 if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
299 if ((ret = vorbis_analysis_wrote(&s->vd, 0)) < 0) {
300 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
301 return vorbis_error_to_averror(ret);
306 /* retrieve available packets from libvorbis */
307 while ((ret = vorbis_analysis_blockout(&s->vd, &s->vb)) == 1) {
308 if ((ret = vorbis_analysis(&s->vb, NULL)) < 0)
310 if ((ret = vorbis_bitrate_addblock(&s->vb)) < 0)
313 /* add any available packets to the output packet buffer */
314 while ((ret = vorbis_bitrate_flushpacket(&s->vd, &op)) == 1) {
315 if (av_fifo_space(s->pkt_fifo) < sizeof(ogg_packet) + op.bytes) {
316 av_log(avctx, AV_LOG_ERROR, "packet buffer is too small\n");
319 av_fifo_generic_write(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
320 av_fifo_generic_write(s->pkt_fifo, op.packet, op.bytes, NULL);
323 av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
328 av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
329 return vorbis_error_to_averror(ret);
332 /* check for available packets */
333 if (av_fifo_size(s->pkt_fifo) < sizeof(ogg_packet))
336 av_fifo_generic_read(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
338 if ((ret = ff_alloc_packet2(avctx, avpkt, op.bytes)) < 0)
340 av_fifo_generic_read(s->pkt_fifo, avpkt->data, op.bytes, NULL);
342 avpkt->pts = ff_samples_to_time_base(avctx, op.granulepos);
344 duration = avpriv_vorbis_parse_frame(&s->vp, avpkt->data, avpkt->size);
346 /* we do not know encoder delay until we get the first packet from
347 * libvorbis, so we have to update the AudioFrameQueue counts */
348 if (!avctx->delay && s->afq.frames) {
349 avctx->delay = duration;
350 av_assert0(!s->afq.remaining_delay);
351 s->afq.frames->duration += duration;
352 s->afq.frames->pts -= duration;
353 s->afq.remaining_samples += duration;
355 ff_af_queue_remove(&s->afq, duration, &avpkt->pts, &avpkt->duration);
362 AVCodec ff_libvorbis_encoder = {
364 .long_name = NULL_IF_CONFIG_SMALL("libvorbis"),
365 .type = AVMEDIA_TYPE_AUDIO,
366 .id = AV_CODEC_ID_VORBIS,
367 .priv_data_size = sizeof(OggVorbisEncContext),
368 .init = oggvorbis_encode_init,
369 .encode2 = oggvorbis_encode_frame,
370 .close = oggvorbis_encode_close,
371 .capabilities = CODEC_CAP_DELAY,
372 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
373 AV_SAMPLE_FMT_NONE },
374 .priv_class = &vorbis_class,
375 .defaults = defaults,