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 * Vorbis encoding 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"
38 /* Number of samples the user should send in each call.
39 * This value is used because it is the LCD of all possible frame sizes, so
40 * an output packet will always start at the same point as one of the input
43 #define OGGVORBIS_FRAME_SIZE 64
45 #define BUFFER_SIZE (1024 * 64)
47 typedef struct OggVorbisContext {
48 AVClass *av_class; /**< class for AVOptions */
49 vorbis_info vi; /**< vorbis_info used during init */
50 vorbis_dsp_state vd; /**< DSP state used for analysis */
51 vorbis_block vb; /**< vorbis_block used for analysis */
52 uint8_t buffer[BUFFER_SIZE]; /**< output packet buffer */
53 int buffer_index; /**< current buffer position */
54 int eof; /**< end-of-file flag */
55 int dsp_initialized; /**< vd has been initialized */
56 vorbis_comment vc; /**< VorbisComment info */
57 ogg_packet op; /**< ogg packet */
58 double iblock; /**< impulse block bias option */
61 static const AVOption options[] = {
62 { "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 },
66 static const AVCodecDefault defaults[] = {
71 static const AVClass class = { "libvorbis", av_default_item_name, options, 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 OggVorbisContext *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)))
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)))
127 /* impulse block bias */
129 if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &s->iblock)))
133 if ((ret = vorbis_encode_setup_init(vi)))
138 return vorbis_error_to_averror(ret);
141 /* How many bytes are needed for a buffer of length 'l' */
142 static int xiph_len(int l)
144 return 1 + l / 255 + l;
147 static av_cold int oggvorbis_encode_close(AVCodecContext *avctx)
149 OggVorbisContext *s = avctx->priv_data;
151 /* notify vorbisenc this is EOF */
152 if (s->dsp_initialized)
153 vorbis_analysis_wrote(&s->vd, 0);
155 vorbis_block_clear(&s->vb);
156 vorbis_dsp_clear(&s->vd);
157 vorbis_info_clear(&s->vi);
159 av_freep(&avctx->coded_frame);
160 av_freep(&avctx->extradata);
165 static av_cold int oggvorbis_encode_init(AVCodecContext *avctx)
167 OggVorbisContext *s = avctx->priv_data;
168 ogg_packet header, header_comm, header_code;
173 vorbis_info_init(&s->vi);
174 if ((ret = oggvorbis_init_encoder(&s->vi, avctx))) {
175 av_log(avctx, AV_LOG_ERROR, "oggvorbis_encode_init: init_encoder failed\n");
178 if ((ret = vorbis_analysis_init(&s->vd, &s->vi))) {
179 ret = vorbis_error_to_averror(ret);
182 s->dsp_initialized = 1;
183 if ((ret = vorbis_block_init(&s->vd, &s->vb))) {
184 ret = vorbis_error_to_averror(ret);
188 vorbis_comment_init(&s->vc);
189 vorbis_comment_add_tag(&s->vc, "encoder", LIBAVCODEC_IDENT);
191 if ((ret = vorbis_analysis_headerout(&s->vd, &s->vc, &header, &header_comm,
193 ret = vorbis_error_to_averror(ret);
197 avctx->extradata_size = 1 + xiph_len(header.bytes) +
198 xiph_len(header_comm.bytes) +
200 p = avctx->extradata = av_malloc(avctx->extradata_size +
201 FF_INPUT_BUFFER_PADDING_SIZE);
203 ret = AVERROR(ENOMEM);
208 offset += av_xiphlacing(&p[offset], header.bytes);
209 offset += av_xiphlacing(&p[offset], header_comm.bytes);
210 memcpy(&p[offset], header.packet, header.bytes);
211 offset += header.bytes;
212 memcpy(&p[offset], header_comm.packet, header_comm.bytes);
213 offset += header_comm.bytes;
214 memcpy(&p[offset], header_code.packet, header_code.bytes);
215 offset += header_code.bytes;
216 assert(offset == avctx->extradata_size);
218 vorbis_comment_clear(&s->vc);
220 avctx->frame_size = OGGVORBIS_FRAME_SIZE;
222 avctx->coded_frame = avcodec_alloc_frame();
223 if (!avctx->coded_frame) {
224 ret = AVERROR(ENOMEM);
230 oggvorbis_encode_close(avctx);
234 static int oggvorbis_encode_frame(AVCodecContext *avctx, unsigned char *packets,
235 int buf_size, void *data)
237 OggVorbisContext *s = avctx->priv_data;
242 /* send samples to libvorbis */
244 const int samples = avctx->frame_size;
246 int c, channels = s->vi.channels;
248 buffer = vorbis_analysis_buffer(&s->vd, samples);
249 for (c = 0; c < channels; c++) {
251 int co = (channels > 8) ? c :
252 ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
253 for (i = 0; i < samples; i++)
254 buffer[c][i] = audio[i * channels + co];
256 if ((ret = vorbis_analysis_wrote(&s->vd, samples)) < 0)
257 return vorbis_error_to_averror(ret);
260 if ((ret = vorbis_analysis_wrote(&s->vd, 0)) < 0)
261 return vorbis_error_to_averror(ret);
265 /* retrieve available packets from libvorbis */
266 while ((ret = vorbis_analysis_blockout(&s->vd, &s->vb)) == 1) {
267 if ((ret = vorbis_analysis(&s->vb, NULL)) < 0)
269 if ((ret = vorbis_bitrate_addblock(&s->vb)) < 0)
272 /* add any available packets to the output packet buffer */
273 while ((ret = vorbis_bitrate_flushpacket(&s->vd, &op)) == 1) {
274 if (s->buffer_index + sizeof(ogg_packet) + op.bytes > BUFFER_SIZE) {
275 av_log(avctx, AV_LOG_ERROR, "libvorbis: buffer overflow.");
278 memcpy(s->buffer + s->buffer_index, &op, sizeof(ogg_packet));
279 s->buffer_index += sizeof(ogg_packet);
280 memcpy(s->buffer + s->buffer_index, op.packet, op.bytes);
281 s->buffer_index += op.bytes;
287 return vorbis_error_to_averror(ret);
289 /* output then next packet from the output buffer, if available */
291 if (s->buffer_index) {
292 ogg_packet *op2 = (ogg_packet *)s->buffer;
293 op2->packet = s->buffer + sizeof(ogg_packet);
295 pkt_size = op2->bytes;
296 // FIXME: we should use the user-supplied pts and duration
297 avctx->coded_frame->pts = ff_samples_to_time_base(avctx,
299 if (pkt_size > buf_size) {
300 av_log(avctx, AV_LOG_ERROR, "libvorbis: buffer overflow.");
304 memcpy(packets, op2->packet, pkt_size);
305 s->buffer_index -= pkt_size + sizeof(ogg_packet);
306 memmove(s->buffer, s->buffer + pkt_size + sizeof(ogg_packet),
313 AVCodec ff_libvorbis_encoder = {
315 .type = AVMEDIA_TYPE_AUDIO,
316 .id = CODEC_ID_VORBIS,
317 .priv_data_size = sizeof(OggVorbisContext),
318 .init = oggvorbis_encode_init,
319 .encode = oggvorbis_encode_frame,
320 .close = oggvorbis_encode_close,
321 .capabilities = CODEC_CAP_DELAY,
322 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLT,
323 AV_SAMPLE_FMT_NONE },
324 .long_name = NULL_IF_CONFIG_SMALL("libvorbis Vorbis"),
325 .priv_class = &class,
326 .defaults = defaults,