2 * Interface to libmp3lame for mp3 encoding
3 * Copyright (c) 2002 Lennert Buytenhek <buytenh@gnu.org>
5 * This file is part of FFmpeg.
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
24 * Interface to libmp3lame for mp3 encoding.
27 #include <lame/lame.h>
29 #include "libavutil/audioconvert.h"
30 #include "libavutil/common.h"
31 #include "libavutil/intreadwrite.h"
32 #include "libavutil/log.h"
33 #include "libavutil/opt.h"
35 #include "audio_frame_queue.h"
37 #include "mpegaudio.h"
38 #include "mpegaudiodecheader.h"
40 #define BUFFER_SIZE (7200 + 2 * MPA_FRAME_SIZE + MPA_FRAME_SIZE / 4+1000) // FIXME: Buffer size to small? Adding 1000 to make up for it.
42 typedef struct LAMEContext {
44 AVCodecContext *avctx;
45 lame_global_flags *gfp;
46 uint8_t buffer[BUFFER_SIZE];
49 void *planar_samples[2];
54 static av_cold int mp3lame_encode_close(AVCodecContext *avctx)
56 LAMEContext *s = avctx->priv_data;
58 #if FF_API_OLD_ENCODE_AUDIO
59 av_freep(&avctx->coded_frame);
61 av_freep(&s->planar_samples[0]);
62 av_freep(&s->planar_samples[1]);
64 ff_af_queue_close(&s->afq);
70 static av_cold int mp3lame_encode_init(AVCodecContext *avctx)
72 LAMEContext *s = avctx->priv_data;
77 /* initialize LAME and get defaults */
78 if ((s->gfp = lame_init()) == NULL)
79 return AVERROR(ENOMEM);
82 lame_set_num_channels(s->gfp, avctx->channels);
83 lame_set_mode(s->gfp, avctx->channels > 1 ? JOINT_STEREO : MONO);
86 lame_set_in_samplerate (s->gfp, avctx->sample_rate);
87 lame_set_out_samplerate(s->gfp, avctx->sample_rate);
89 /* algorithmic quality */
90 if (avctx->compression_level == FF_COMPRESSION_DEFAULT)
91 lame_set_quality(s->gfp, 5);
93 lame_set_quality(s->gfp, avctx->compression_level);
96 if (avctx->flags & CODEC_FLAG_QSCALE) {
97 lame_set_VBR(s->gfp, vbr_default);
98 lame_set_VBR_quality(s->gfp, avctx->global_quality / (float)FF_QP2LAMBDA);
101 lame_set_brate(s->gfp, avctx->bit_rate / 1000);
104 /* do not get a Xing VBR header frame from LAME */
105 lame_set_bWriteVbrTag(s->gfp,0);
107 /* bit reservoir usage */
108 lame_set_disable_reservoir(s->gfp, !s->reservoir);
110 /* set specified parameters */
111 if (lame_init_params(s->gfp) < 0) {
116 /* get encoder delay */
117 avctx->delay = lame_get_encoder_delay(s->gfp) + 528 + 1;
118 ff_af_queue_init(avctx, &s->afq);
120 avctx->frame_size = lame_get_framesize(s->gfp);
122 #if FF_API_OLD_ENCODE_AUDIO
123 avctx->coded_frame = avcodec_alloc_frame();
124 if (!avctx->coded_frame) {
125 ret = AVERROR(ENOMEM);
131 if (avctx->sample_fmt == AV_SAMPLE_FMT_S32 ||
132 avctx->sample_fmt == AV_SAMPLE_FMT_FLT) {
134 for (ch = 0; ch < avctx->channels; ch++) {
135 s->planar_samples[ch] = av_malloc(avctx->frame_size *
136 av_get_bytes_per_sample(avctx->sample_fmt));
137 if (!s->planar_samples[ch]) {
138 ret = AVERROR(ENOMEM);
146 mp3lame_encode_close(avctx);
150 #define DEINTERLEAVE(type, scale) do { \
152 for (ch = 0; ch < s->avctx->channels; ch++) { \
153 const type *input = samples; \
154 type *output = s->planar_samples[ch]; \
156 for (i = 0; i < nb_samples; i++) { \
157 output[i] = *input * scale; \
158 input += s->avctx->channels; \
163 static int encode_frame_int16(LAMEContext *s, void *samples, int nb_samples)
165 if (s->avctx->channels > 1) {
166 return lame_encode_buffer_interleaved(s->gfp, samples,
168 s->buffer + s->buffer_index,
169 BUFFER_SIZE - s->buffer_index);
171 return lame_encode_buffer(s->gfp, samples, NULL, nb_samples,
172 s->buffer + s->buffer_index,
173 BUFFER_SIZE - s->buffer_index);
177 static int encode_frame_int32(LAMEContext *s, void *samples, int nb_samples)
179 DEINTERLEAVE(int32_t, 1);
181 return lame_encode_buffer_int(s->gfp,
182 s->planar_samples[0], s->planar_samples[1],
184 s->buffer + s->buffer_index,
185 BUFFER_SIZE - s->buffer_index);
188 static int encode_frame_float(LAMEContext *s, void *samples, int nb_samples)
190 DEINTERLEAVE(float, 32768.0f);
192 return lame_encode_buffer_float(s->gfp,
193 s->planar_samples[0], s->planar_samples[1],
195 s->buffer + s->buffer_index,
196 BUFFER_SIZE - s->buffer_index);
199 static int mp3lame_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
200 const AVFrame *frame, int *got_packet_ptr)
202 LAMEContext *s = avctx->priv_data;
208 switch (avctx->sample_fmt) {
209 case AV_SAMPLE_FMT_S16:
210 lame_result = encode_frame_int16(s, frame->data[0], frame->nb_samples);
212 case AV_SAMPLE_FMT_S32:
213 lame_result = encode_frame_int32(s, frame->data[0], frame->nb_samples);
215 case AV_SAMPLE_FMT_FLT:
216 lame_result = encode_frame_float(s, frame->data[0], frame->nb_samples);
222 lame_result = lame_encode_flush(s->gfp, s->buffer + s->buffer_index,
223 BUFFER_SIZE - s->buffer_index);
225 if (lame_result < 0) {
226 if (lame_result == -1) {
227 av_log(avctx, AV_LOG_ERROR,
228 "lame: output buffer too small (buffer index: %d, free bytes: %d)\n",
229 s->buffer_index, BUFFER_SIZE - s->buffer_index);
233 s->buffer_index += lame_result;
235 /* add current frame to the queue */
237 if ((ret = ff_af_queue_add(&s->afq, frame) < 0))
241 /* Move 1 frame from the LAME buffer to the output packet, if available.
242 We have to parse the first frame header in the output buffer to
243 determine the frame size. */
244 if (s->buffer_index < 4)
246 if (avpriv_mpegaudio_decode_header(&hdr, AV_RB32(s->buffer))) {
247 av_log(avctx, AV_LOG_ERROR, "free format output not supported\n");
250 len = hdr.frame_size;
251 av_dlog(avctx, "in:%d packet-len:%d index:%d\n", avctx->frame_size, len,
253 if (len <= s->buffer_index) {
254 if ((ret = ff_alloc_packet2(avctx, avpkt, len)))
256 memcpy(avpkt->data, s->buffer, len);
257 s->buffer_index -= len;
258 memmove(s->buffer, s->buffer + len, s->buffer_index);
260 /* Get the next frame pts/duration */
261 ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
270 #define OFFSET(x) offsetof(LAMEContext, x)
271 #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
272 static const AVOption options[] = {
273 { "reservoir", "Use bit reservoir.", OFFSET(reservoir), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, AE },
277 static const AVClass libmp3lame_class = {
278 .class_name = "libmp3lame encoder",
279 .item_name = av_default_item_name,
281 .version = LIBAVUTIL_VERSION_INT,
284 static const AVCodecDefault libmp3lame_defaults[] = {
289 static const int libmp3lame_sample_rates[] = {
290 44100, 48000, 32000, 22050, 24000, 16000, 11025, 12000, 8000, 0
293 AVCodec ff_libmp3lame_encoder = {
294 .name = "libmp3lame",
295 .type = AVMEDIA_TYPE_AUDIO,
296 .id = AV_CODEC_ID_MP3,
297 .priv_data_size = sizeof(LAMEContext),
298 .init = mp3lame_encode_init,
299 .encode2 = mp3lame_encode_frame,
300 .close = mp3lame_encode_close,
301 .capabilities = CODEC_CAP_DELAY | CODEC_CAP_SMALL_LAST_FRAME,
302 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S32,
305 AV_SAMPLE_FMT_NONE },
306 .supported_samplerates = libmp3lame_sample_rates,
307 .channel_layouts = (const uint64_t[]) { AV_CH_LAYOUT_MONO,
310 .long_name = NULL_IF_CONFIG_SMALL("libmp3lame MP3 (MPEG audio layer 3)"),
311 .priv_class = &libmp3lame_class,
312 .defaults = libmp3lame_defaults,