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/fifo.h"
24 #include "libavutil/opt.h"
26 #include "audio_frame_queue.h"
29 #include "vorbis_parser.h"
34 /* Number of samples the user should send in each call.
35 * This value is used because it is the LCD of all possible frame sizes, so
36 * an output packet will always start at the same point as one of the input
39 #define OGGVORBIS_FRAME_SIZE 64
41 #define BUFFER_SIZE (1024 * 64)
43 typedef struct OggVorbisEncContext {
44 AVClass *av_class; /**< class for AVOptions */
46 vorbis_info vi; /**< vorbis_info used during init */
47 vorbis_dsp_state vd; /**< DSP state used for analysis */
48 vorbis_block vb; /**< vorbis_block used for analysis */
49 AVFifoBuffer *pkt_fifo; /**< output packet buffer */
50 int eof; /**< end-of-file flag */
51 int dsp_initialized; /**< vd has been initialized */
52 vorbis_comment vc; /**< VorbisComment info */
53 double iblock; /**< impulse block bias option */
54 VorbisParseContext vp; /**< parse context to get durations */
55 AudioFrameQueue afq; /**< frame queue for timestamps */
56 } OggVorbisEncContext;
58 static const AVOption options[] = {
59 { "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 },
63 static const AVCodecDefault defaults[] = {
68 static const AVClass class = { "libvorbis", av_default_item_name, options, LIBAVUTIL_VERSION_INT };
71 static int vorbis_error_to_averror(int ov_err)
74 case OV_EFAULT: return AVERROR_BUG;
75 case OV_EINVAL: return AVERROR(EINVAL);
76 case OV_EIMPL: return AVERROR(EINVAL);
77 default: return AVERROR_UNKNOWN;
81 static av_cold int oggvorbis_init_encoder(vorbis_info *vi,
82 AVCodecContext *avctx)
84 OggVorbisEncContext *s = avctx->priv_data;
88 if (avctx->flags & CODEC_FLAG_QSCALE || !avctx->bit_rate) {
90 * NOTE: we use the oggenc range of -1 to 10 for global_quality for
91 * user convenience, but libvorbis uses -0.1 to 1.0.
93 float q = avctx->global_quality / (float)FF_QP2LAMBDA;
94 /* default to 3 if the user did not set quality or bitrate */
95 if (!(avctx->flags & CODEC_FLAG_QSCALE))
97 if ((ret = vorbis_encode_setup_vbr(vi, avctx->channels,
102 int minrate = avctx->rc_min_rate > 0 ? avctx->rc_min_rate : -1;
103 int maxrate = avctx->rc_max_rate > 0 ? avctx->rc_max_rate : -1;
105 /* average bitrate */
106 if ((ret = vorbis_encode_setup_managed(vi, avctx->channels,
107 avctx->sample_rate, maxrate,
108 avctx->bit_rate, minrate)))
111 /* variable bitrate by estimate, disable slow rate management */
112 if (minrate == -1 && maxrate == -1)
113 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL)))
114 goto error; /* should not happen */
117 /* cutoff frequency */
118 if (avctx->cutoff > 0) {
119 cfreq = avctx->cutoff / 1000.0;
120 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq)))
121 goto error; /* should not happen */
124 /* impulse block bias */
126 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &s->iblock)))
130 if (avctx->channels == 3 &&
131 avctx->channel_layout != (AV_CH_LAYOUT_STEREO|AV_CH_FRONT_CENTER) ||
132 avctx->channels == 4 &&
133 avctx->channel_layout != AV_CH_LAYOUT_2_2 &&
134 avctx->channel_layout != AV_CH_LAYOUT_QUAD ||
135 avctx->channels == 5 &&
136 avctx->channel_layout != AV_CH_LAYOUT_5POINT0 &&
137 avctx->channel_layout != AV_CH_LAYOUT_5POINT0_BACK ||
138 avctx->channels == 6 &&
139 avctx->channel_layout != AV_CH_LAYOUT_5POINT1 &&
140 avctx->channel_layout != AV_CH_LAYOUT_5POINT1_BACK ||
141 avctx->channels == 7 &&
142 avctx->channel_layout != (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_CENTER) ||
143 avctx->channels == 8 &&
144 avctx->channel_layout != AV_CH_LAYOUT_7POINT1) {
145 if (avctx->channel_layout) {
147 av_get_channel_layout_string(name, sizeof(name), avctx->channels,
148 avctx->channel_layout);
149 av_log(avctx, AV_LOG_ERROR, "%s not supported by Vorbis: "
150 "output stream will have incorrect "
151 "channel layout.\n", name);
153 av_log(avctx, AV_LOG_WARNING, "No channel layout specified. The encoder "
154 "will use Vorbis channel layout for "
155 "%d channels.\n", avctx->channels);
159 if ((ret = vorbis_encode_setup_init(vi)))
164 return vorbis_error_to_averror(ret);
167 /* How many bytes are needed for a buffer of length 'l' */
168 static int xiph_len(int l)
170 return 1 + l / 255 + l;
173 static av_cold int oggvorbis_encode_close(AVCodecContext *avctx)
175 OggVorbisEncContext *s = avctx->priv_data;
177 /* notify vorbisenc this is EOF */
178 if (s->dsp_initialized)
179 vorbis_analysis_wrote(&s->vd, 0);
181 vorbis_block_clear(&s->vb);
182 vorbis_dsp_clear(&s->vd);
183 vorbis_info_clear(&s->vi);
185 av_fifo_free(s->pkt_fifo);
186 ff_af_queue_close(&s->afq);
187 #if FF_API_OLD_ENCODE_AUDIO
188 av_freep(&avctx->coded_frame);
190 av_freep(&avctx->extradata);
195 static av_cold int oggvorbis_encode_init(AVCodecContext *avctx)
197 OggVorbisEncContext *s = avctx->priv_data;
198 ogg_packet header, header_comm, header_code;
203 vorbis_info_init(&s->vi);
204 if ((ret = oggvorbis_init_encoder(&s->vi, avctx))) {
205 av_log(avctx, AV_LOG_ERROR, "encoder setup failed\n");
208 if ((ret = vorbis_analysis_init(&s->vd, &s->vi))) {
209 av_log(avctx, AV_LOG_ERROR, "analysis init failed\n");
210 ret = vorbis_error_to_averror(ret);
213 s->dsp_initialized = 1;
214 if ((ret = vorbis_block_init(&s->vd, &s->vb))) {
215 av_log(avctx, AV_LOG_ERROR, "dsp init failed\n");
216 ret = vorbis_error_to_averror(ret);
220 vorbis_comment_init(&s->vc);
221 if (!(avctx->flags & CODEC_FLAG_BITEXACT))
222 vorbis_comment_add_tag(&s->vc, "encoder", LIBAVCODEC_IDENT);
224 if ((ret = vorbis_analysis_headerout(&s->vd, &s->vc, &header, &header_comm,
226 ret = vorbis_error_to_averror(ret);
230 avctx->extradata_size = 1 + xiph_len(header.bytes) +
231 xiph_len(header_comm.bytes) +
233 p = avctx->extradata = av_malloc(avctx->extradata_size +
234 FF_INPUT_BUFFER_PADDING_SIZE);
236 ret = AVERROR(ENOMEM);
241 offset += av_xiphlacing(&p[offset], header.bytes);
242 offset += av_xiphlacing(&p[offset], header_comm.bytes);
243 memcpy(&p[offset], header.packet, header.bytes);
244 offset += header.bytes;
245 memcpy(&p[offset], header_comm.packet, header_comm.bytes);
246 offset += header_comm.bytes;
247 memcpy(&p[offset], header_code.packet, header_code.bytes);
248 offset += header_code.bytes;
249 assert(offset == avctx->extradata_size);
251 if ((ret = avpriv_vorbis_parse_extradata(avctx, &s->vp)) < 0) {
252 av_log(avctx, AV_LOG_ERROR, "invalid extradata\n");
256 vorbis_comment_clear(&s->vc);
258 avctx->frame_size = OGGVORBIS_FRAME_SIZE;
259 ff_af_queue_init(avctx, &s->afq);
261 s->pkt_fifo = av_fifo_alloc(BUFFER_SIZE);
263 ret = AVERROR(ENOMEM);
267 #if FF_API_OLD_ENCODE_AUDIO
268 avctx->coded_frame = avcodec_alloc_frame();
269 if (!avctx->coded_frame) {
270 ret = AVERROR(ENOMEM);
277 oggvorbis_encode_close(avctx);
281 static int oggvorbis_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
282 const AVFrame *frame, int *got_packet_ptr)
284 OggVorbisEncContext *s = avctx->priv_data;
288 /* send samples to libvorbis */
290 const float *audio = (const float *)frame->data[0];
291 const int samples = frame->nb_samples;
293 int c, channels = s->vi.channels;
295 buffer = vorbis_analysis_buffer(&s->vd, samples);
296 for (c = 0; c < channels; c++) {
298 int co = (channels > 8) ? c :
299 ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
300 for (i = 0; i < samples; i++)
301 buffer[c][i] = audio[i * channels + co];
303 if ((ret = vorbis_analysis_wrote(&s->vd, samples)) < 0) {
304 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
305 return vorbis_error_to_averror(ret);
307 if ((ret = ff_af_queue_add(&s->afq, frame) < 0))
311 if ((ret = vorbis_analysis_wrote(&s->vd, 0)) < 0) {
312 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
313 return vorbis_error_to_averror(ret);
318 /* retrieve available packets from libvorbis */
319 while ((ret = vorbis_analysis_blockout(&s->vd, &s->vb)) == 1) {
320 if ((ret = vorbis_analysis(&s->vb, NULL)) < 0)
322 if ((ret = vorbis_bitrate_addblock(&s->vb)) < 0)
325 /* add any available packets to the output packet buffer */
326 while ((ret = vorbis_bitrate_flushpacket(&s->vd, &op)) == 1) {
327 if (av_fifo_space(s->pkt_fifo) < sizeof(ogg_packet) + op.bytes) {
328 av_log(avctx, AV_LOG_ERROR, "packet buffer is too small\n");
331 av_fifo_generic_write(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
332 av_fifo_generic_write(s->pkt_fifo, op.packet, op.bytes, NULL);
335 av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
340 av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
341 return vorbis_error_to_averror(ret);
344 /* check for available packets */
345 if (av_fifo_size(s->pkt_fifo) < sizeof(ogg_packet))
348 av_fifo_generic_read(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
350 if ((ret = ff_alloc_packet2(avctx, avpkt, op.bytes)))
352 av_fifo_generic_read(s->pkt_fifo, avpkt->data, op.bytes, NULL);
354 avpkt->pts = ff_samples_to_time_base(avctx, op.granulepos);
356 duration = avpriv_vorbis_parse_frame(&s->vp, avpkt->data, avpkt->size);
358 /* we do not know encoder delay until we get the first packet from
359 * libvorbis, so we have to update the AudioFrameQueue counts */
361 avctx->delay = duration;
362 s->afq.remaining_delay += duration;
363 s->afq.remaining_samples += duration;
365 ff_af_queue_remove(&s->afq, duration, &avpkt->pts, &avpkt->duration);
372 AVCodec ff_libvorbis_encoder = {
374 .type = AVMEDIA_TYPE_AUDIO,
375 .id = CODEC_ID_VORBIS,
376 .priv_data_size = sizeof(OggVorbisEncContext),
377 .init = oggvorbis_encode_init,
378 .encode2 = oggvorbis_encode_frame,
379 .close = oggvorbis_encode_close,
380 .capabilities = CODEC_CAP_DELAY,
381 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLT,
382 AV_SAMPLE_FMT_NONE },
383 .long_name = NULL_IF_CONFIG_SMALL("libvorbis"),
384 .priv_class = &class,
385 .defaults = defaults,