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 St, 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>
30 #include "bytestream.h"
35 #define OGGVORBIS_FRAME_SIZE 64
37 #define BUFFER_SIZE (1024*64)
39 typedef struct OggVorbisContext {
43 uint8_t buffer[BUFFER_SIZE];
52 static int oggvorbis_init_encoder(vorbis_info *vi, AVCodecContext *avccontext) {
55 if(avccontext->flags & CODEC_FLAG_QSCALE) {
56 /* variable bitrate */
57 if(vorbis_encode_setup_vbr(vi, avccontext->channels,
58 avccontext->sample_rate,
59 avccontext->global_quality / (float)FF_QP2LAMBDA))
62 /* constant bitrate */
63 if(vorbis_encode_setup_managed(vi, avccontext->channels,
64 avccontext->sample_rate, -1, avccontext->bit_rate, -1))
67 #ifdef OGGVORBIS_VBR_BY_ESTIMATE
68 /* variable bitrate by estimate */
69 if(vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE_AVG, NULL))
74 /* cutoff frequency */
75 if(avccontext->cutoff > 0) {
76 cfreq = avccontext->cutoff / 1000.0;
77 if(vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq))
81 return vorbis_encode_setup_init(vi);
84 static int oggvorbis_encode_init(AVCodecContext *avccontext) {
85 OggVorbisContext *context = avccontext->priv_data ;
86 ogg_packet header, header_comm, header_code;
88 unsigned int offset, len;
90 vorbis_info_init(&context->vi) ;
91 if(oggvorbis_init_encoder(&context->vi, avccontext) < 0) {
92 av_log(avccontext, AV_LOG_ERROR, "oggvorbis_encode_init: init_encoder failed") ;
95 vorbis_analysis_init(&context->vd, &context->vi) ;
96 vorbis_block_init(&context->vd, &context->vb) ;
98 vorbis_comment_init(&context->vc);
99 vorbis_comment_add_tag(&context->vc, "encoder", LIBAVCODEC_IDENT) ;
101 vorbis_analysis_headerout(&context->vd, &context->vc, &header,
102 &header_comm, &header_code);
104 len = header.bytes + header_comm.bytes + header_code.bytes;
105 avccontext->extradata_size= 64 + len + len/255;
106 p = avccontext->extradata= av_mallocz(avccontext->extradata_size);
109 offset += av_xiphlacing(&p[offset], header.bytes);
110 offset += av_xiphlacing(&p[offset], header_comm.bytes);
111 memcpy(&p[offset], header.packet, header.bytes);
112 offset += header.bytes;
113 memcpy(&p[offset], header_comm.packet, header_comm.bytes);
114 offset += header_comm.bytes;
115 memcpy(&p[offset], header_code.packet, header_code.bytes);
116 offset += header_code.bytes;
117 avccontext->extradata_size = offset;
118 avccontext->extradata= av_realloc(avccontext->extradata, avccontext->extradata_size);
120 /* vorbis_block_clear(&context->vb);
121 vorbis_dsp_clear(&context->vd);
122 vorbis_info_clear(&context->vi);*/
123 vorbis_comment_clear(&context->vc);
125 avccontext->frame_size = OGGVORBIS_FRAME_SIZE ;
127 avccontext->coded_frame= avcodec_alloc_frame();
128 avccontext->coded_frame->key_frame= 1;
134 static int oggvorbis_encode_frame(AVCodecContext *avccontext,
135 unsigned char *packets,
136 int buf_size, void *data)
138 OggVorbisContext *context = avccontext->priv_data ;
141 signed short *audio = data ;
142 int l, samples = data ? OGGVORBIS_FRAME_SIZE : 0;
144 buffer = vorbis_analysis_buffer(&context->vd, samples) ;
146 if(context->vi.channels == 1) {
147 for(l = 0 ; l < samples ; l++)
148 buffer[0][l]=audio[l]/32768.f;
150 for(l = 0 ; l < samples ; l++){
151 buffer[0][l]=audio[l*2]/32768.f;
152 buffer[1][l]=audio[l*2+1]/32768.f;
156 vorbis_analysis_wrote(&context->vd, samples) ;
158 while(vorbis_analysis_blockout(&context->vd, &context->vb) == 1) {
159 vorbis_analysis(&context->vb, NULL);
160 vorbis_bitrate_addblock(&context->vb) ;
162 while(vorbis_bitrate_flushpacket(&context->vd, &op)) {
163 /* i'd love to say the following line is a hack, but sadly it's
164 * not, apparently the end of stream decision is in libogg. */
167 memcpy(context->buffer + context->buffer_index, &op, sizeof(ogg_packet));
168 context->buffer_index += sizeof(ogg_packet);
169 memcpy(context->buffer + context->buffer_index, op.packet, op.bytes);
170 context->buffer_index += op.bytes;
171 // av_log(avccontext, AV_LOG_DEBUG, "e%d / %d\n", context->buffer_index, op.bytes);
176 if(context->buffer_index){
177 ogg_packet *op2= (ogg_packet*)context->buffer;
178 op2->packet = context->buffer + sizeof(ogg_packet);
181 avccontext->coded_frame->pts= av_rescale_q(op2->granulepos, (AVRational){1, avccontext->sample_rate}, avccontext->time_base);
182 //FIXME we should reorder the user supplied pts and not assume that they are spaced by 1/sample_rate
184 memcpy(packets, op2->packet, l);
185 context->buffer_index -= l + sizeof(ogg_packet);
186 memcpy(context->buffer, context->buffer + l + sizeof(ogg_packet), context->buffer_index);
187 // av_log(avccontext, AV_LOG_DEBUG, "E%d\n", l);
194 static int oggvorbis_encode_close(AVCodecContext *avccontext) {
195 OggVorbisContext *context = avccontext->priv_data ;
196 /* ogg_packet op ; */
198 vorbis_analysis_wrote(&context->vd, 0) ; /* notify vorbisenc this is EOF */
200 vorbis_block_clear(&context->vb);
201 vorbis_dsp_clear(&context->vd);
202 vorbis_info_clear(&context->vi);
204 av_freep(&avccontext->coded_frame);
205 av_freep(&avccontext->extradata);
211 AVCodec libvorbis_encoder = {
215 sizeof(OggVorbisContext),
216 oggvorbis_encode_init,
217 oggvorbis_encode_frame,
218 oggvorbis_encode_close,
219 .capabilities= CODEC_CAP_DELAY,
222 static int oggvorbis_decode_init(AVCodecContext *avccontext) {
223 OggVorbisContext *context = avccontext->priv_data ;
224 uint8_t *p= avccontext->extradata;
226 unsigned char *headers[3], *extradata = avccontext->extradata;
228 vorbis_info_init(&context->vi) ;
229 vorbis_comment_init(&context->vc) ;
231 if(! avccontext->extradata_size || ! p) {
232 av_log(avccontext, AV_LOG_ERROR, "vorbis extradata absent\n");
236 if(p[0] == 0 && p[1] == 30) {
237 for(i = 0; i < 3; i++){
238 hsizes[i] = bytestream_get_be16(&p);
243 unsigned int offset = 1;
247 while((*p == 0xFF) && (offset < avccontext->extradata_size)) {
252 if(offset >= avccontext->extradata_size - 1) {
253 av_log(avccontext, AV_LOG_ERROR,
254 "vorbis header sizes damaged\n");
261 hsizes[2] = avccontext->extradata_size - hsizes[0]-hsizes[1]-offset;
263 av_log(avccontext, AV_LOG_DEBUG,
264 "vorbis header sizes: %d, %d, %d, / extradata_len is %d \n",
265 hsizes[0], hsizes[1], hsizes[2], avccontext->extradata_size);
267 headers[0] = extradata + offset;
268 headers[1] = extradata + offset + hsizes[0];
269 headers[2] = extradata + offset + hsizes[0] + hsizes[1];
271 av_log(avccontext, AV_LOG_ERROR,
272 "vorbis initial header len is wrong: %d\n", *p);
277 context->op.b_o_s= i==0;
278 context->op.bytes = hsizes[i];
279 context->op.packet = headers[i];
280 if(vorbis_synthesis_headerin(&context->vi, &context->vc, &context->op)<0){
281 av_log(avccontext, AV_LOG_ERROR, "%d. vorbis header damaged\n", i+1);
286 avccontext->channels = context->vi.channels;
287 avccontext->sample_rate = context->vi.rate;
288 avccontext->time_base= (AVRational){1, avccontext->sample_rate};
290 vorbis_synthesis_init(&context->vd, &context->vi);
291 vorbis_block_init(&context->vd, &context->vb);
297 static inline int conv(int samples, float **pcm, char *buf, int channels) {
299 ogg_int16_t *ptr, *data = (ogg_int16_t*)buf ;
302 for(i = 0 ; i < channels ; i++){
306 for(j = 0 ; j < samples ; j++) {
308 val = mono[j] * 32767.f;
310 if(val > 32767) val = 32767 ;
311 if(val < -32768) val = -32768 ;
322 static int oggvorbis_decode_frame(AVCodecContext *avccontext,
323 void *data, int *data_size,
324 uint8_t *buf, int buf_size)
326 OggVorbisContext *context = avccontext->priv_data ;
328 ogg_packet *op= &context->op;
329 int samples, total_samples, total_bytes;
337 op->bytes = buf_size;
339 // av_log(avccontext, AV_LOG_DEBUG, "%d %d %d %"PRId64" %"PRId64" %d %d\n", op->bytes, op->b_o_s, op->e_o_s, op->granulepos, op->packetno, buf_size, context->vi.rate);
341 /* for(i=0; i<op->bytes; i++)
342 av_log(avccontext, AV_LOG_DEBUG, "%02X ", op->packet[i]);
343 av_log(avccontext, AV_LOG_DEBUG, "\n");*/
345 if(vorbis_synthesis(&context->vb, op) == 0)
346 vorbis_synthesis_blockin(&context->vd, &context->vb) ;
351 while((samples = vorbis_synthesis_pcmout(&context->vd, &pcm)) > 0) {
352 conv(samples, pcm, (char*)data + total_bytes, context->vi.channels) ;
353 total_bytes += samples * 2 * context->vi.channels ;
354 total_samples += samples ;
355 vorbis_synthesis_read(&context->vd, samples) ;
358 *data_size = total_bytes ;
363 static int oggvorbis_decode_close(AVCodecContext *avccontext) {
364 OggVorbisContext *context = avccontext->priv_data ;
366 vorbis_info_clear(&context->vi) ;
367 vorbis_comment_clear(&context->vc) ;
373 AVCodec libvorbis_decoder = {
377 sizeof(OggVorbisContext),
378 oggvorbis_decode_init,
380 oggvorbis_decode_close,
381 oggvorbis_decode_frame,
382 .capabilities= CODEC_CAP_DELAY,