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/intreadwrite.h"
30 #include "libavutil/log.h"
31 #include "libavutil/opt.h"
34 #include "mpegaudio.h"
35 #include "mpegaudiodecheader.h"
37 #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.
39 typedef struct LAMEContext {
41 AVCodecContext *avctx;
42 lame_global_flags *gfp;
43 uint8_t buffer[BUFFER_SIZE];
46 void *planar_samples[2];
50 static av_cold int mp3lame_encode_close(AVCodecContext *avctx)
52 LAMEContext *s = avctx->priv_data;
54 av_freep(&avctx->coded_frame);
55 av_freep(&s->planar_samples[0]);
56 av_freep(&s->planar_samples[1]);
62 static av_cold int mp3lame_encode_init(AVCodecContext *avctx)
64 LAMEContext *s = avctx->priv_data;
69 /* initialize LAME and get defaults */
70 if ((s->gfp = lame_init()) == NULL)
71 return AVERROR(ENOMEM);
74 if (avctx->channels > 2) {
75 av_log(avctx, AV_LOG_ERROR,
76 "Invalid number of channels %d, must be <= 2\n", avctx->channels);
77 ret = AVERROR(EINVAL);
80 lame_set_num_channels(s->gfp, avctx->channels);
81 lame_set_mode(s->gfp, avctx->channels > 1 ? JOINT_STEREO : MONO);
84 lame_set_in_samplerate (s->gfp, avctx->sample_rate);
85 lame_set_out_samplerate(s->gfp, avctx->sample_rate);
87 /* algorithmic quality */
88 if (avctx->compression_level == FF_COMPRESSION_DEFAULT)
89 lame_set_quality(s->gfp, 5);
91 lame_set_quality(s->gfp, avctx->compression_level);
94 if (avctx->flags & CODEC_FLAG_QSCALE) {
95 lame_set_VBR(s->gfp, vbr_default);
96 lame_set_VBR_quality(s->gfp, avctx->global_quality / (float)FF_QP2LAMBDA);
99 lame_set_brate(s->gfp, avctx->bit_rate / 1000);
102 /* do not get a Xing VBR header frame from LAME */
103 lame_set_bWriteVbrTag(s->gfp,0);
105 /* bit reservoir usage */
106 lame_set_disable_reservoir(s->gfp, !s->reservoir);
108 /* set specified parameters */
109 if (lame_init_params(s->gfp) < 0) {
114 avctx->frame_size = lame_get_framesize(s->gfp);
115 avctx->coded_frame = avcodec_alloc_frame();
116 if (!avctx->coded_frame) {
117 ret = AVERROR(ENOMEM);
122 if (avctx->sample_fmt == AV_SAMPLE_FMT_S32 ||
123 avctx->sample_fmt == AV_SAMPLE_FMT_FLT) {
125 for (ch = 0; ch < avctx->channels; ch++) {
126 s->planar_samples[ch] = av_malloc(avctx->frame_size *
127 av_get_bytes_per_sample(avctx->sample_fmt));
128 if (!s->planar_samples[ch]) {
129 ret = AVERROR(ENOMEM);
137 mp3lame_encode_close(avctx);
141 #define DEINTERLEAVE(type, scale) do { \
143 for (ch = 0; ch < s->avctx->channels; ch++) { \
144 const type *input = samples; \
145 type *output = s->planar_samples[ch]; \
147 for (i = 0; i < s->avctx->frame_size; i++) { \
148 output[i] = *input * scale; \
149 input += s->avctx->channels; \
154 static int encode_frame_int16(LAMEContext *s, void *samples)
156 if (s->avctx->channels > 1) {
157 return lame_encode_buffer_interleaved(s->gfp, samples,
158 s->avctx->frame_size,
159 s->buffer + s->buffer_index,
160 BUFFER_SIZE - s->buffer_index);
162 return lame_encode_buffer(s->gfp, samples, NULL, s->avctx->frame_size,
163 s->buffer + s->buffer_index,
164 BUFFER_SIZE - s->buffer_index);
168 static int encode_frame_int32(LAMEContext *s, void *samples)
170 DEINTERLEAVE(int32_t, 1);
172 return lame_encode_buffer_int(s->gfp,
173 s->planar_samples[0], s->planar_samples[1],
174 s->avctx->frame_size,
175 s->buffer + s->buffer_index,
176 BUFFER_SIZE - s->buffer_index);
179 static int encode_frame_float(LAMEContext *s, void *samples)
181 DEINTERLEAVE(float, 32768.0f);
183 return lame_encode_buffer_float(s->gfp,
184 s->planar_samples[0], s->planar_samples[1],
185 s->avctx->frame_size,
186 s->buffer + s->buffer_index,
187 BUFFER_SIZE - s->buffer_index);
190 static int mp3lame_encode_frame(AVCodecContext *avctx, unsigned char *frame,
191 int buf_size, void *data)
193 LAMEContext *s = avctx->priv_data;
199 switch (avctx->sample_fmt) {
200 case AV_SAMPLE_FMT_S16:
201 lame_result = encode_frame_int16(s, data);
203 case AV_SAMPLE_FMT_S32:
204 lame_result = encode_frame_int32(s, data);
206 case AV_SAMPLE_FMT_FLT:
207 lame_result = encode_frame_float(s, data);
213 lame_result = lame_encode_flush(s->gfp, s->buffer + s->buffer_index,
214 BUFFER_SIZE - s->buffer_index);
216 if (lame_result < 0) {
217 if (lame_result == -1) {
218 av_log(avctx, AV_LOG_ERROR,
219 "lame: output buffer too small (buffer index: %d, free bytes: %d)\n",
220 s->buffer_index, BUFFER_SIZE - s->buffer_index);
224 s->buffer_index += lame_result;
226 /* Move 1 frame from the LAME buffer to the output packet, if available.
227 We have to parse the first frame header in the output buffer to
228 determine the frame size. */
229 if (s->buffer_index < 4)
231 if (avpriv_mpegaudio_decode_header(&hdr, AV_RB32(s->buffer))) {
232 av_log(avctx, AV_LOG_ERROR, "free format output not supported\n");
235 len = hdr.frame_size;
236 av_dlog(avctx, "in:%d packet-len:%d index:%d\n", avctx->frame_size, len,
238 if (len <= s->buffer_index) {
239 memcpy(frame, s->buffer, len);
240 s->buffer_index -= len;
241 memmove(s->buffer, s->buffer + len, s->buffer_index);
247 #define OFFSET(x) offsetof(LAMEContext, x)
248 #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
249 static const AVOption options[] = {
250 { "reservoir", "Use bit reservoir.", OFFSET(reservoir), AV_OPT_TYPE_INT, { 1 }, 0, 1, AE },
254 static const AVClass libmp3lame_class = {
255 .class_name = "libmp3lame encoder",
256 .item_name = av_default_item_name,
258 .version = LIBAVUTIL_VERSION_INT,
261 static const AVCodecDefault libmp3lame_defaults[] = {
266 static const int libmp3lame_sample_rates[] = {
267 44100, 48000, 32000, 22050, 24000, 16000, 11025, 12000, 8000, 0
270 AVCodec ff_libmp3lame_encoder = {
271 .name = "libmp3lame",
272 .type = AVMEDIA_TYPE_AUDIO,
274 .priv_data_size = sizeof(LAMEContext),
275 .init = mp3lame_encode_init,
276 .encode = mp3lame_encode_frame,
277 .close = mp3lame_encode_close,
278 .capabilities = CODEC_CAP_DELAY,
279 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S32,
282 AV_SAMPLE_FMT_NONE },
283 .supported_samplerates = libmp3lame_sample_rates,
284 .long_name = NULL_IF_CONFIG_SMALL("libmp3lame MP3 (MPEG audio layer 3)"),
285 .priv_class = &libmp3lame_class,
286 .defaults = libmp3lame_defaults,