ee76ff88128fc04362110b5b6217cc4ec3e170f5
[libav.git] / libavcodec / libmp3lame.c
1 /*
2 * Interface to libmp3lame for mp3 encoding
3 * Copyright (c) 2002 Lennert Buytenhek <buytenh@gnu.org>
4 *
5 * This file is part of Libav.
6 *
7 * Libav 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.
11 *
12 * Libav 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.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22 /**
23 * @file
24 * Interface to libmp3lame for mp3 encoding.
25 */
26
27 #include <lame/lame.h>
28
29 #include "libavutil/channel_layout.h"
30 #include "libavutil/common.h"
31 #include "libavutil/float_dsp.h"
32 #include "libavutil/intreadwrite.h"
33 #include "libavutil/log.h"
34 #include "libavutil/opt.h"
35 #include "avcodec.h"
36 #include "audio_frame_queue.h"
37 #include "internal.h"
38 #include "mpegaudio.h"
39 #include "mpegaudiodecheader.h"
40
41 #define BUFFER_SIZE (7200 + 2 * MPA_FRAME_SIZE + MPA_FRAME_SIZE / 4)
42
43 typedef struct LAMEContext {
44 AVClass *class;
45 AVCodecContext *avctx;
46 lame_global_flags *gfp;
47 uint8_t *buffer;
48 int buffer_index;
49 int buffer_size;
50 int reservoir;
51 float *samples_flt[2];
52 AudioFrameQueue afq;
53 AVFloatDSPContext fdsp;
54 } LAMEContext;
55
56
57 static int realloc_buffer(LAMEContext *s)
58 {
59 if (!s->buffer || s->buffer_size - s->buffer_index < BUFFER_SIZE) {
60 int new_size = s->buffer_index + 2 * BUFFER_SIZE, err;
61
62 av_dlog(s->avctx, "resizing output buffer: %d -> %d\n", s->buffer_size,
63 new_size);
64 if ((err = av_reallocp(&s->buffer, new_size)) < 0) {
65 s->buffer_size = s->buffer_index = 0;
66 return err;
67 }
68 s->buffer_size = new_size;
69 }
70 return 0;
71 }
72
73 static av_cold int mp3lame_encode_close(AVCodecContext *avctx)
74 {
75 LAMEContext *s = avctx->priv_data;
76
77 av_freep(&s->samples_flt[0]);
78 av_freep(&s->samples_flt[1]);
79 av_freep(&s->buffer);
80
81 ff_af_queue_close(&s->afq);
82
83 lame_close(s->gfp);
84 return 0;
85 }
86
87 static av_cold int mp3lame_encode_init(AVCodecContext *avctx)
88 {
89 LAMEContext *s = avctx->priv_data;
90 int ret;
91
92 s->avctx = avctx;
93
94 /* initialize LAME and get defaults */
95 if ((s->gfp = lame_init()) == NULL)
96 return AVERROR(ENOMEM);
97
98 lame_set_num_channels(s->gfp, avctx->channels);
99 lame_set_mode(s->gfp, avctx->channels > 1 ? JOINT_STEREO : MONO);
100
101 /* sample rate */
102 lame_set_in_samplerate (s->gfp, avctx->sample_rate);
103 lame_set_out_samplerate(s->gfp, avctx->sample_rate);
104
105 /* algorithmic quality */
106 if (avctx->compression_level == FF_COMPRESSION_DEFAULT)
107 lame_set_quality(s->gfp, 5);
108 else
109 lame_set_quality(s->gfp, avctx->compression_level);
110
111 /* rate control */
112 if (avctx->flags & CODEC_FLAG_QSCALE) {
113 lame_set_VBR(s->gfp, vbr_default);
114 lame_set_VBR_quality(s->gfp, avctx->global_quality / (float)FF_QP2LAMBDA);
115 } else {
116 if (avctx->bit_rate)
117 lame_set_brate(s->gfp, avctx->bit_rate / 1000);
118 }
119
120 /* do not get a Xing VBR header frame from LAME */
121 lame_set_bWriteVbrTag(s->gfp,0);
122
123 /* bit reservoir usage */
124 lame_set_disable_reservoir(s->gfp, !s->reservoir);
125
126 /* set specified parameters */
127 if (lame_init_params(s->gfp) < 0) {
128 ret = -1;
129 goto error;
130 }
131
132 /* get encoder delay */
133 avctx->delay = lame_get_encoder_delay(s->gfp) + 528 + 1;
134 ff_af_queue_init(avctx, &s->afq);
135
136 avctx->frame_size = lame_get_framesize(s->gfp);
137
138 /* allocate float sample buffers */
139 if (avctx->sample_fmt == AV_SAMPLE_FMT_FLTP) {
140 int ch;
141 for (ch = 0; ch < avctx->channels; ch++) {
142 s->samples_flt[ch] = av_malloc(avctx->frame_size *
143 sizeof(*s->samples_flt[ch]));
144 if (!s->samples_flt[ch]) {
145 ret = AVERROR(ENOMEM);
146 goto error;
147 }
148 }
149 }
150
151 ret = realloc_buffer(s);
152 if (ret < 0)
153 goto error;
154
155 avpriv_float_dsp_init(&s->fdsp, avctx->flags & CODEC_FLAG_BITEXACT);
156
157 return 0;
158 error:
159 mp3lame_encode_close(avctx);
160 return ret;
161 }
162
163 #define ENCODE_BUFFER(func, buf_type, buf_name) do { \
164 lame_result = func(s->gfp, \
165 (const buf_type *)buf_name[0], \
166 (const buf_type *)buf_name[1], frame->nb_samples, \
167 s->buffer + s->buffer_index, \
168 s->buffer_size - s->buffer_index); \
169 } while (0)
170
171 static int mp3lame_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
172 const AVFrame *frame, int *got_packet_ptr)
173 {
174 LAMEContext *s = avctx->priv_data;
175 MPADecodeHeader hdr;
176 int len, ret, ch;
177 int lame_result;
178
179 if (frame) {
180 switch (avctx->sample_fmt) {
181 case AV_SAMPLE_FMT_S16P:
182 ENCODE_BUFFER(lame_encode_buffer, int16_t, frame->data);
183 break;
184 case AV_SAMPLE_FMT_S32P:
185 ENCODE_BUFFER(lame_encode_buffer_int, int32_t, frame->data);
186 break;
187 case AV_SAMPLE_FMT_FLTP:
188 if (frame->linesize[0] < 4 * FFALIGN(frame->nb_samples, 8)) {
189 av_log(avctx, AV_LOG_ERROR, "inadequate AVFrame plane padding\n");
190 return AVERROR(EINVAL);
191 }
192 for (ch = 0; ch < avctx->channels; ch++) {
193 s->fdsp.vector_fmul_scalar(s->samples_flt[ch],
194 (const float *)frame->data[ch],
195 32768.0f,
196 FFALIGN(frame->nb_samples, 8));
197 }
198 ENCODE_BUFFER(lame_encode_buffer_float, float, s->samples_flt);
199 break;
200 default:
201 return AVERROR_BUG;
202 }
203 } else {
204 lame_result = lame_encode_flush(s->gfp, s->buffer + s->buffer_index,
205 s->buffer_size - s->buffer_index);
206 }
207 if (lame_result < 0) {
208 if (lame_result == -1) {
209 av_log(avctx, AV_LOG_ERROR,
210 "lame: output buffer too small (buffer index: %d, free bytes: %d)\n",
211 s->buffer_index, s->buffer_size - s->buffer_index);
212 }
213 return -1;
214 }
215 s->buffer_index += lame_result;
216 ret = realloc_buffer(s);
217 if (ret < 0) {
218 av_log(avctx, AV_LOG_ERROR, "error reallocating output buffer\n");
219 return ret;
220 }
221
222 /* add current frame to the queue */
223 if (frame) {
224 if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
225 return ret;
226 }
227
228 /* Move 1 frame from the LAME buffer to the output packet, if available.
229 We have to parse the first frame header in the output buffer to
230 determine the frame size. */
231 if (s->buffer_index < 4)
232 return 0;
233 if (avpriv_mpegaudio_decode_header(&hdr, AV_RB32(s->buffer))) {
234 av_log(avctx, AV_LOG_ERROR, "free format output not supported\n");
235 return -1;
236 }
237 len = hdr.frame_size;
238 av_dlog(avctx, "in:%d packet-len:%d index:%d\n", avctx->frame_size, len,
239 s->buffer_index);
240 if (len <= s->buffer_index) {
241 if ((ret = ff_alloc_packet(avpkt, len))) {
242 av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
243 return ret;
244 }
245 memcpy(avpkt->data, s->buffer, len);
246 s->buffer_index -= len;
247 memmove(s->buffer, s->buffer + len, s->buffer_index);
248
249 /* Get the next frame pts/duration */
250 ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
251 &avpkt->duration);
252
253 avpkt->size = len;
254 *got_packet_ptr = 1;
255 }
256 return 0;
257 }
258
259 #define OFFSET(x) offsetof(LAMEContext, x)
260 #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
261 static const AVOption options[] = {
262 { "reservoir", "Use bit reservoir.", OFFSET(reservoir), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, AE },
263 { NULL },
264 };
265
266 static const AVClass libmp3lame_class = {
267 .class_name = "libmp3lame encoder",
268 .item_name = av_default_item_name,
269 .option = options,
270 .version = LIBAVUTIL_VERSION_INT,
271 };
272
273 static const AVCodecDefault libmp3lame_defaults[] = {
274 { "b", "0" },
275 { NULL },
276 };
277
278 static const int libmp3lame_sample_rates[] = {
279 44100, 48000, 32000, 22050, 24000, 16000, 11025, 12000, 8000, 0
280 };
281
282 AVCodec ff_libmp3lame_encoder = {
283 .name = "libmp3lame",
284 .long_name = NULL_IF_CONFIG_SMALL("libmp3lame MP3 (MPEG audio layer 3)"),
285 .type = AVMEDIA_TYPE_AUDIO,
286 .id = AV_CODEC_ID_MP3,
287 .priv_data_size = sizeof(LAMEContext),
288 .init = mp3lame_encode_init,
289 .encode2 = mp3lame_encode_frame,
290 .close = mp3lame_encode_close,
291 .capabilities = CODEC_CAP_DELAY | CODEC_CAP_SMALL_LAST_FRAME,
292 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S32P,
293 AV_SAMPLE_FMT_FLTP,
294 AV_SAMPLE_FMT_S16P,
295 AV_SAMPLE_FMT_NONE },
296 .supported_samplerates = libmp3lame_sample_rates,
297 .channel_layouts = (const uint64_t[]) { AV_CH_LAYOUT_MONO,
298 AV_CH_LAYOUT_STEREO,
299 0 },
300 .priv_class = &libmp3lame_class,
301 .defaults = libmp3lame_defaults,
302 };