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/intreadwrite.h"
31 #include "libavutil/log.h"
32 #include "libavutil/opt.h"
34 #include "audio_frame_queue.h"
36 #include "mpegaudio.h"
37 #include "mpegaudiodecheader.h"
39 #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.
41 typedef struct LAMEContext {
43 AVCodecContext *avctx;
44 lame_global_flags *gfp;
45 uint8_t buffer[BUFFER_SIZE];
48 void *planar_samples[2];
53 static av_cold int mp3lame_encode_close(AVCodecContext *avctx)
55 LAMEContext *s = avctx->priv_data;
57 #if FF_API_OLD_ENCODE_AUDIO
58 av_freep(&avctx->coded_frame);
60 av_freep(&s->planar_samples[0]);
61 av_freep(&s->planar_samples[1]);
63 ff_af_queue_close(&s->afq);
69 static av_cold int mp3lame_encode_init(AVCodecContext *avctx)
71 LAMEContext *s = avctx->priv_data;
76 /* initialize LAME and get defaults */
77 if ((s->gfp = lame_init()) == NULL)
78 return AVERROR(ENOMEM);
81 lame_set_num_channels(s->gfp, avctx->channels);
82 lame_set_mode(s->gfp, avctx->channels > 1 ? JOINT_STEREO : MONO);
85 lame_set_in_samplerate (s->gfp, avctx->sample_rate);
86 lame_set_out_samplerate(s->gfp, avctx->sample_rate);
88 /* algorithmic quality */
89 if (avctx->compression_level == FF_COMPRESSION_DEFAULT)
90 lame_set_quality(s->gfp, 5);
92 lame_set_quality(s->gfp, avctx->compression_level);
95 if (avctx->flags & CODEC_FLAG_QSCALE) {
96 lame_set_VBR(s->gfp, vbr_default);
97 lame_set_VBR_quality(s->gfp, avctx->global_quality / (float)FF_QP2LAMBDA);
100 lame_set_brate(s->gfp, avctx->bit_rate / 1000);
103 /* do not get a Xing VBR header frame from LAME */
104 lame_set_bWriteVbrTag(s->gfp,0);
106 /* bit reservoir usage */
107 lame_set_disable_reservoir(s->gfp, !s->reservoir);
109 /* set specified parameters */
110 if (lame_init_params(s->gfp) < 0) {
115 /* get encoder delay */
116 avctx->delay = lame_get_encoder_delay(s->gfp) + 528 + 1;
117 ff_af_queue_init(avctx, &s->afq);
119 avctx->frame_size = lame_get_framesize(s->gfp);
121 #if FF_API_OLD_ENCODE_AUDIO
122 avctx->coded_frame = avcodec_alloc_frame();
123 if (!avctx->coded_frame) {
124 ret = AVERROR(ENOMEM);
130 if (avctx->sample_fmt == AV_SAMPLE_FMT_S32 ||
131 avctx->sample_fmt == AV_SAMPLE_FMT_FLT) {
133 for (ch = 0; ch < avctx->channels; ch++) {
134 s->planar_samples[ch] = av_malloc(avctx->frame_size *
135 av_get_bytes_per_sample(avctx->sample_fmt));
136 if (!s->planar_samples[ch]) {
137 ret = AVERROR(ENOMEM);
145 mp3lame_encode_close(avctx);
149 #define DEINTERLEAVE(type, scale) do { \
151 for (ch = 0; ch < s->avctx->channels; ch++) { \
152 const type *input = samples; \
153 type *output = s->planar_samples[ch]; \
155 for (i = 0; i < nb_samples; i++) { \
156 output[i] = *input * scale; \
157 input += s->avctx->channels; \
162 static int encode_frame_int16(LAMEContext *s, void *samples, int nb_samples)
164 if (s->avctx->channels > 1) {
165 return lame_encode_buffer_interleaved(s->gfp, samples,
167 s->buffer + s->buffer_index,
168 BUFFER_SIZE - s->buffer_index);
170 return lame_encode_buffer(s->gfp, samples, NULL, nb_samples,
171 s->buffer + s->buffer_index,
172 BUFFER_SIZE - s->buffer_index);
176 static int encode_frame_int32(LAMEContext *s, void *samples, int nb_samples)
178 DEINTERLEAVE(int32_t, 1);
180 return lame_encode_buffer_int(s->gfp,
181 s->planar_samples[0], s->planar_samples[1],
183 s->buffer + s->buffer_index,
184 BUFFER_SIZE - s->buffer_index);
187 static int encode_frame_float(LAMEContext *s, void *samples, int nb_samples)
189 DEINTERLEAVE(float, 32768.0f);
191 return lame_encode_buffer_float(s->gfp,
192 s->planar_samples[0], s->planar_samples[1],
194 s->buffer + s->buffer_index,
195 BUFFER_SIZE - s->buffer_index);
198 static int mp3lame_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
199 const AVFrame *frame, int *got_packet_ptr)
201 LAMEContext *s = avctx->priv_data;
207 switch (avctx->sample_fmt) {
208 case AV_SAMPLE_FMT_S16:
209 lame_result = encode_frame_int16(s, frame->data[0], frame->nb_samples);
211 case AV_SAMPLE_FMT_S32:
212 lame_result = encode_frame_int32(s, frame->data[0], frame->nb_samples);
214 case AV_SAMPLE_FMT_FLT:
215 lame_result = encode_frame_float(s, frame->data[0], frame->nb_samples);
221 lame_result = lame_encode_flush(s->gfp, s->buffer + s->buffer_index,
222 BUFFER_SIZE - s->buffer_index);
224 if (lame_result < 0) {
225 if (lame_result == -1) {
226 av_log(avctx, AV_LOG_ERROR,
227 "lame: output buffer too small (buffer index: %d, free bytes: %d)\n",
228 s->buffer_index, BUFFER_SIZE - s->buffer_index);
232 s->buffer_index += lame_result;
234 /* add current frame to the queue */
236 if ((ret = ff_af_queue_add(&s->afq, frame) < 0))
240 /* Move 1 frame from the LAME buffer to the output packet, if available.
241 We have to parse the first frame header in the output buffer to
242 determine the frame size. */
243 if (s->buffer_index < 4)
245 if (avpriv_mpegaudio_decode_header(&hdr, AV_RB32(s->buffer))) {
246 av_log(avctx, AV_LOG_ERROR, "free format output not supported\n");
249 len = hdr.frame_size;
250 av_dlog(avctx, "in:%d packet-len:%d index:%d\n", avctx->frame_size, len,
252 if (len <= s->buffer_index) {
253 if ((ret = ff_alloc_packet2(avctx, avpkt, len)))
255 memcpy(avpkt->data, s->buffer, len);
256 s->buffer_index -= len;
257 memmove(s->buffer, s->buffer + len, s->buffer_index);
259 /* Get the next frame pts/duration */
260 ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
269 #define OFFSET(x) offsetof(LAMEContext, x)
270 #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
271 static const AVOption options[] = {
272 { "reservoir", "Use bit reservoir.", OFFSET(reservoir), AV_OPT_TYPE_INT, { 1 }, 0, 1, AE },
276 static const AVClass libmp3lame_class = {
277 .class_name = "libmp3lame encoder",
278 .item_name = av_default_item_name,
280 .version = LIBAVUTIL_VERSION_INT,
283 static const AVCodecDefault libmp3lame_defaults[] = {
288 static const int libmp3lame_sample_rates[] = {
289 44100, 48000, 32000, 22050, 24000, 16000, 11025, 12000, 8000, 0
292 AVCodec ff_libmp3lame_encoder = {
293 .name = "libmp3lame",
294 .type = AVMEDIA_TYPE_AUDIO,
296 .priv_data_size = sizeof(LAMEContext),
297 .init = mp3lame_encode_init,
298 .encode2 = mp3lame_encode_frame,
299 .close = mp3lame_encode_close,
300 .capabilities = CODEC_CAP_DELAY | CODEC_CAP_SMALL_LAST_FRAME,
301 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S32,
304 AV_SAMPLE_FMT_NONE },
305 .supported_samplerates = libmp3lame_sample_rates,
306 .channel_layouts = (const uint64_t[]) { AV_CH_LAYOUT_MONO,
309 .long_name = NULL_IF_CONFIG_SMALL("libmp3lame MP3 (MPEG audio layer 3)"),
310 .priv_class = &libmp3lame_class,
311 .defaults = libmp3lame_defaults,