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 * Ogg Vorbis codec support via libvorbisenc.
24 * @author Mark Hills <mark@pogo.org.uk>
27 #include <vorbis/vorbisenc.h>
29 #include "libavutil/opt.h"
31 #include "bytestream.h"
33 #include "libavutil/mathematics.h"
38 #define OGGVORBIS_FRAME_SIZE 64
40 #define BUFFER_SIZE (1024 * 64)
42 typedef struct OggVorbisContext {
47 uint8_t buffer[BUFFER_SIZE];
58 static const AVOption options[] = {
59 { "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 },
62 static const AVClass class = { "libvorbis", av_default_item_name, options, LIBAVUTIL_VERSION_INT };
64 static const char * error(int oggerr, int *averr)
67 case OV_EFAULT: *averr = AVERROR(EFAULT); return "internal error";
68 case OV_EIMPL: *averr = AVERROR(EINVAL); return "not supported";
69 case OV_EINVAL: *averr = AVERROR(EINVAL); return "invalid request";
70 default: *averr = AVERROR(EINVAL); return "unknown error";
74 static av_cold int oggvorbis_init_encoder(vorbis_info *vi, AVCodecContext *avccontext)
76 OggVorbisContext *context = avccontext->priv_data;
80 if (avccontext->flags & CODEC_FLAG_QSCALE) {
81 /* variable bitrate */
82 float quality = avccontext->global_quality / (float)FF_QP2LAMBDA;
83 r = vorbis_encode_setup_vbr(vi, avccontext->channels,
84 avccontext->sample_rate,
87 av_log(avccontext, AV_LOG_ERROR,
88 "Unable to set quality to %g: %s\n", quality, error(r, &r));
92 int minrate = avccontext->rc_min_rate > 0 ? avccontext->rc_min_rate : -1;
93 int maxrate = avccontext->rc_min_rate > 0 ? avccontext->rc_max_rate : -1;
95 /* constant bitrate */
96 r = vorbis_encode_setup_managed(vi, avccontext->channels,
97 avccontext->sample_rate, minrate,
98 avccontext->bit_rate, maxrate);
100 av_log(avccontext, AV_LOG_ERROR,
101 "Unable to set CBR to %d: %s\n", avccontext->bit_rate,
106 /* variable bitrate by estimate, disable slow rate management */
107 if (minrate == -1 && maxrate == -1)
108 if (vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL))
109 return AVERROR(EINVAL); /* should not happen */
112 /* cutoff frequency */
113 if (avccontext->cutoff > 0) {
114 cfreq = avccontext->cutoff / 1000.0;
115 if (vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq))
116 return AVERROR(EINVAL); /* should not happen */
119 if (context->iblock) {
120 vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &context->iblock);
123 if (avccontext->channels == 3 &&
124 avccontext->channel_layout != (AV_CH_LAYOUT_STEREO|AV_CH_FRONT_CENTER) ||
125 avccontext->channels == 4 &&
126 avccontext->channel_layout != AV_CH_LAYOUT_2_2 &&
127 avccontext->channel_layout != AV_CH_LAYOUT_QUAD ||
128 avccontext->channels == 5 &&
129 avccontext->channel_layout != AV_CH_LAYOUT_5POINT0 &&
130 avccontext->channel_layout != AV_CH_LAYOUT_5POINT0_BACK ||
131 avccontext->channels == 6 &&
132 avccontext->channel_layout != AV_CH_LAYOUT_5POINT1 &&
133 avccontext->channel_layout != AV_CH_LAYOUT_5POINT1_BACK ||
134 avccontext->channels == 7 &&
135 avccontext->channel_layout != (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_CENTER) ||
136 avccontext->channels == 8 &&
137 avccontext->channel_layout != AV_CH_LAYOUT_7POINT1) {
138 if (avccontext->channel_layout) {
140 av_get_channel_layout_string(name, sizeof(name), avccontext->channels,
141 avccontext->channel_layout);
142 av_log(avccontext, AV_LOG_ERROR, "%s not supported by Vorbis: "
143 "output stream will have incorrect "
144 "channel layout.\n", name);
146 av_log(avccontext, AV_LOG_WARNING, "No channel layout specified. The encoder "
147 "will use Vorbis channel layout for "
148 "%d channels.\n", avccontext->channels);
152 return vorbis_encode_setup_init(vi);
155 /* How many bytes are needed for a buffer of length 'l' */
156 static int xiph_len(int l)
158 return 1 + l / 255 + l;
161 static av_cold int oggvorbis_encode_init(AVCodecContext *avccontext)
163 OggVorbisContext *context = avccontext->priv_data;
164 ogg_packet header, header_comm, header_code;
169 vorbis_info_init(&context->vi);
170 r = oggvorbis_init_encoder(&context->vi, avccontext);
172 av_log(avccontext, AV_LOG_ERROR, "oggvorbis_encode_init failed\n");
175 vorbis_analysis_init(&context->vd, &context->vi);
176 vorbis_block_init(&context->vd, &context->vb);
178 vorbis_comment_init(&context->vc);
179 vorbis_comment_add_tag(&context->vc, "encoder", LIBAVCODEC_IDENT);
181 vorbis_analysis_headerout(&context->vd, &context->vc, &header,
182 &header_comm, &header_code);
184 avccontext->extradata_size =
185 1 + xiph_len(header.bytes) + xiph_len(header_comm.bytes) +
187 p = avccontext->extradata =
188 av_malloc(avccontext->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
191 offset += av_xiphlacing(&p[offset], header.bytes);
192 offset += av_xiphlacing(&p[offset], header_comm.bytes);
193 memcpy(&p[offset], header.packet, header.bytes);
194 offset += header.bytes;
195 memcpy(&p[offset], header_comm.packet, header_comm.bytes);
196 offset += header_comm.bytes;
197 memcpy(&p[offset], header_code.packet, header_code.bytes);
198 offset += header_code.bytes;
199 assert(offset == avccontext->extradata_size);
202 vorbis_block_clear(&context->vb);
203 vorbis_dsp_clear(&context->vd);
204 vorbis_info_clear(&context->vi);
206 vorbis_comment_clear(&context->vc);
208 avccontext->frame_size = OGGVORBIS_FRAME_SIZE;
210 avccontext->coded_frame = avcodec_alloc_frame();
211 avccontext->coded_frame->key_frame = 1;
216 static int oggvorbis_encode_frame(AVCodecContext *avccontext,
217 unsigned char *packets,
218 int buf_size, void *data)
220 OggVorbisContext *context = avccontext->priv_data;
222 signed short *audio = data;
226 const int samples = avccontext->frame_size;
228 int c, channels = context->vi.channels;
230 buffer = vorbis_analysis_buffer(&context->vd, samples);
231 for (c = 0; c < channels; c++) {
232 int co = (channels > 8) ? c :
233 ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
234 for (l = 0; l < samples; l++)
235 buffer[c][l] = audio[l * channels + co] / 32768.f;
237 vorbis_analysis_wrote(&context->vd, samples);
240 vorbis_analysis_wrote(&context->vd, 0);
244 while (vorbis_analysis_blockout(&context->vd, &context->vb) == 1) {
245 vorbis_analysis(&context->vb, NULL);
246 vorbis_bitrate_addblock(&context->vb);
248 while (vorbis_bitrate_flushpacket(&context->vd, &op)) {
249 /* i'd love to say the following line is a hack, but sadly it's
250 * not, apparently the end of stream decision is in libogg. */
251 if (op.bytes == 1 && op.e_o_s)
253 if (context->buffer_index + sizeof(ogg_packet) + op.bytes > BUFFER_SIZE) {
254 av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.\n");
255 return AVERROR(EINVAL);
257 memcpy(context->buffer + context->buffer_index, &op, sizeof(ogg_packet));
258 context->buffer_index += sizeof(ogg_packet);
259 memcpy(context->buffer + context->buffer_index, op.packet, op.bytes);
260 context->buffer_index += op.bytes;
261 // av_log(avccontext, AV_LOG_DEBUG, "e%d / %d\n", context->buffer_index, op.bytes);
266 if (context->buffer_index) {
267 ogg_packet *op2 = (ogg_packet *)context->buffer;
268 op2->packet = context->buffer + sizeof(ogg_packet);
271 avccontext->coded_frame->pts = av_rescale_q(op2->granulepos, (AVRational) { 1, avccontext->sample_rate }, avccontext->time_base);
272 //FIXME we should reorder the user supplied pts and not assume that they are spaced by 1/sample_rate
275 av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.\n");
276 return AVERROR(EINVAL);
279 memcpy(packets, op2->packet, l);
280 context->buffer_index -= l + sizeof(ogg_packet);
281 memmove(context->buffer, context->buffer + l + sizeof(ogg_packet), context->buffer_index);
282 // av_log(avccontext, AV_LOG_DEBUG, "E%d\n", l);
288 static av_cold int oggvorbis_encode_close(AVCodecContext *avccontext)
290 OggVorbisContext *context = avccontext->priv_data;
291 /* ogg_packet op ; */
293 vorbis_analysis_wrote(&context->vd, 0); /* notify vorbisenc this is EOF */
295 vorbis_block_clear(&context->vb);
296 vorbis_dsp_clear(&context->vd);
297 vorbis_info_clear(&context->vi);
299 av_freep(&avccontext->coded_frame);
300 av_freep(&avccontext->extradata);
305 AVCodec ff_libvorbis_encoder = {
307 .type = AVMEDIA_TYPE_AUDIO,
308 .id = CODEC_ID_VORBIS,
309 .priv_data_size = sizeof(OggVorbisContext),
310 .init = oggvorbis_encode_init,
311 .encode = oggvorbis_encode_frame,
312 .close = oggvorbis_encode_close,
313 .capabilities = CODEC_CAP_DELAY,
314 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE },
315 .long_name = NULL_IF_CONFIG_SMALL("libvorbis Vorbis"),
316 .priv_class = &class,