2 * copyright (c) 2002 Mark Hills <mark@pogo.org.uk>
4 * This file is part of Libav.
6 * Libav 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 * Libav 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 Libav; 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 av_cold int oggvorbis_init_encoder(vorbis_info *vi, AVCodecContext *avccontext)
66 OggVorbisContext *context = avccontext->priv_data;
69 if (avccontext->flags & CODEC_FLAG_QSCALE) {
70 /* variable bitrate */
71 if (vorbis_encode_setup_vbr(vi, avccontext->channels,
72 avccontext->sample_rate,
73 avccontext->global_quality / (float)FF_QP2LAMBDA / 10.0))
76 int minrate = avccontext->rc_min_rate > 0 ? avccontext->rc_min_rate : -1;
77 int maxrate = avccontext->rc_min_rate > 0 ? avccontext->rc_max_rate : -1;
79 /* constant bitrate */
80 if (vorbis_encode_setup_managed(vi, avccontext->channels,
81 avccontext->sample_rate, minrate,
82 avccontext->bit_rate, maxrate))
85 /* variable bitrate by estimate, disable slow rate management */
86 if (minrate == -1 && maxrate == -1)
87 if (vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL))
91 /* cutoff frequency */
92 if (avccontext->cutoff > 0) {
93 cfreq = avccontext->cutoff / 1000.0;
94 if (vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq))
98 if (context->iblock) {
99 vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &context->iblock);
102 return vorbis_encode_setup_init(vi);
105 /* How many bytes are needed for a buffer of length 'l' */
106 static int xiph_len(int l)
108 return (1 + l / 255 + l);
111 static av_cold int oggvorbis_encode_init(AVCodecContext *avccontext)
113 OggVorbisContext *context = avccontext->priv_data;
114 ogg_packet header, header_comm, header_code;
118 vorbis_info_init(&context->vi);
119 if (oggvorbis_init_encoder(&context->vi, avccontext) < 0) {
120 av_log(avccontext, AV_LOG_ERROR, "oggvorbis_encode_init: init_encoder failed\n");
123 vorbis_analysis_init(&context->vd, &context->vi);
124 vorbis_block_init(&context->vd, &context->vb);
126 vorbis_comment_init(&context->vc);
127 vorbis_comment_add_tag(&context->vc, "encoder", LIBAVCODEC_IDENT);
129 vorbis_analysis_headerout(&context->vd, &context->vc, &header,
130 &header_comm, &header_code);
132 avccontext->extradata_size =
133 1 + xiph_len(header.bytes) + xiph_len(header_comm.bytes) +
135 p = avccontext->extradata =
136 av_malloc(avccontext->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
139 offset += av_xiphlacing(&p[offset], header.bytes);
140 offset += av_xiphlacing(&p[offset], header_comm.bytes);
141 memcpy(&p[offset], header.packet, header.bytes);
142 offset += header.bytes;
143 memcpy(&p[offset], header_comm.packet, header_comm.bytes);
144 offset += header_comm.bytes;
145 memcpy(&p[offset], header_code.packet, header_code.bytes);
146 offset += header_code.bytes;
147 assert(offset == avccontext->extradata_size);
150 vorbis_block_clear(&context->vb);
151 vorbis_dsp_clear(&context->vd);
152 vorbis_info_clear(&context->vi);
154 vorbis_comment_clear(&context->vc);
156 avccontext->frame_size = OGGVORBIS_FRAME_SIZE;
158 avccontext->coded_frame = avcodec_alloc_frame();
159 avccontext->coded_frame->key_frame = 1;
164 static int oggvorbis_encode_frame(AVCodecContext *avccontext,
165 unsigned char *packets,
166 int buf_size, void *data)
168 OggVorbisContext *context = avccontext->priv_data;
170 signed short *audio = data;
174 const int samples = avccontext->frame_size;
176 int c, channels = context->vi.channels;
178 buffer = vorbis_analysis_buffer(&context->vd, samples);
179 for (c = 0; c < channels; c++) {
180 int co = (channels > 8) ? c :
181 ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
182 for (l = 0; l < samples; l++)
183 buffer[c][l] = audio[l * channels + co] / 32768.f;
185 vorbis_analysis_wrote(&context->vd, samples);
188 vorbis_analysis_wrote(&context->vd, 0);
192 while (vorbis_analysis_blockout(&context->vd, &context->vb) == 1) {
193 vorbis_analysis(&context->vb, NULL);
194 vorbis_bitrate_addblock(&context->vb);
196 while (vorbis_bitrate_flushpacket(&context->vd, &op)) {
197 /* i'd love to say the following line is a hack, but sadly it's
198 * not, apparently the end of stream decision is in libogg. */
199 if (op.bytes == 1 && op.e_o_s)
201 if (context->buffer_index + sizeof(ogg_packet) + op.bytes > BUFFER_SIZE) {
202 av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.");
205 memcpy(context->buffer + context->buffer_index, &op, sizeof(ogg_packet));
206 context->buffer_index += sizeof(ogg_packet);
207 memcpy(context->buffer + context->buffer_index, op.packet, op.bytes);
208 context->buffer_index += op.bytes;
209 // av_log(avccontext, AV_LOG_DEBUG, "e%d / %d\n", context->buffer_index, op.bytes);
214 if (context->buffer_index) {
215 ogg_packet *op2 = (ogg_packet *)context->buffer;
216 op2->packet = context->buffer + sizeof(ogg_packet);
219 avccontext->coded_frame->pts = av_rescale_q(op2->granulepos, (AVRational) { 1, avccontext->sample_rate }, avccontext->time_base);
220 //FIXME we should reorder the user supplied pts and not assume that they are spaced by 1/sample_rate
223 av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.");
227 memcpy(packets, op2->packet, l);
228 context->buffer_index -= l + sizeof(ogg_packet);
229 memmove(context->buffer, context->buffer + l + sizeof(ogg_packet), context->buffer_index);
230 // av_log(avccontext, AV_LOG_DEBUG, "E%d\n", l);
236 static av_cold int oggvorbis_encode_close(AVCodecContext *avccontext)
238 OggVorbisContext *context = avccontext->priv_data;
239 /* ogg_packet op ; */
241 vorbis_analysis_wrote(&context->vd, 0); /* notify vorbisenc this is EOF */
243 vorbis_block_clear(&context->vb);
244 vorbis_dsp_clear(&context->vd);
245 vorbis_info_clear(&context->vi);
247 av_freep(&avccontext->coded_frame);
248 av_freep(&avccontext->extradata);
253 AVCodec ff_libvorbis_encoder = {
255 .type = AVMEDIA_TYPE_AUDIO,
256 .id = CODEC_ID_VORBIS,
257 .priv_data_size = sizeof(OggVorbisContext),
258 .init = oggvorbis_encode_init,
259 .encode = oggvorbis_encode_frame,
260 .close = oggvorbis_encode_close,
261 .capabilities = CODEC_CAP_DELAY,
262 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE },
263 .long_name = NULL_IF_CONFIG_SMALL("libvorbis Vorbis"),
264 .priv_class = &class,