libmp3lame: allow joint stereo to be disabled
[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 int joint_stereo;
52 float *samples_flt[2];
53 AudioFrameQueue afq;
54 AVFloatDSPContext fdsp;
55 } LAMEContext;
56
57
58 static int realloc_buffer(LAMEContext *s)
59 {
60 if (!s->buffer || s->buffer_size - s->buffer_index < BUFFER_SIZE) {
61 int new_size = s->buffer_index + 2 * BUFFER_SIZE, err;
62
63 av_dlog(s->avctx, "resizing output buffer: %d -> %d\n", s->buffer_size,
64 new_size);
65 if ((err = av_reallocp(&s->buffer, new_size)) < 0) {
66 s->buffer_size = s->buffer_index = 0;
67 return err;
68 }
69 s->buffer_size = new_size;
70 }
71 return 0;
72 }
73
74 static av_cold int mp3lame_encode_close(AVCodecContext *avctx)
75 {
76 LAMEContext *s = avctx->priv_data;
77
78 av_freep(&s->samples_flt[0]);
79 av_freep(&s->samples_flt[1]);
80 av_freep(&s->buffer);
81
82 ff_af_queue_close(&s->afq);
83
84 lame_close(s->gfp);
85 return 0;
86 }
87
88 static av_cold int mp3lame_encode_init(AVCodecContext *avctx)
89 {
90 LAMEContext *s = avctx->priv_data;
91 int ret;
92
93 s->avctx = avctx;
94
95 /* initialize LAME and get defaults */
96 if ((s->gfp = lame_init()) == NULL)
97 return AVERROR(ENOMEM);
98
99 lame_set_num_channels(s->gfp, avctx->channels);
100 lame_set_mode(s->gfp, avctx->channels > 1 ? s->joint_stereo ? JOINT_STEREO : STEREO : MONO);
101
102 /* sample rate */
103 lame_set_in_samplerate (s->gfp, avctx->sample_rate);
104 lame_set_out_samplerate(s->gfp, avctx->sample_rate);
105
106 /* algorithmic quality */
107 if (avctx->compression_level == FF_COMPRESSION_DEFAULT)
108 lame_set_quality(s->gfp, 5);
109 else
110 lame_set_quality(s->gfp, avctx->compression_level);
111
112 /* rate control */
113 if (avctx->flags & CODEC_FLAG_QSCALE) {
114 lame_set_VBR(s->gfp, vbr_default);
115 lame_set_VBR_quality(s->gfp, avctx->global_quality / (float)FF_QP2LAMBDA);
116 } else {
117 if (avctx->bit_rate)
118 lame_set_brate(s->gfp, avctx->bit_rate / 1000);
119 }
120
121 /* do not get a Xing VBR header frame from LAME */
122 lame_set_bWriteVbrTag(s->gfp,0);
123
124 /* bit reservoir usage */
125 lame_set_disable_reservoir(s->gfp, !s->reservoir);
126
127 /* set specified parameters */
128 if (lame_init_params(s->gfp) < 0) {
129 ret = -1;
130 goto error;
131 }
132
133 /* get encoder delay */
134 avctx->delay = lame_get_encoder_delay(s->gfp) + 528 + 1;
135 ff_af_queue_init(avctx, &s->afq);
136
137 avctx->frame_size = lame_get_framesize(s->gfp);
138
139 /* allocate float sample buffers */
140 if (avctx->sample_fmt == AV_SAMPLE_FMT_FLTP) {
141 int ch;
142 for (ch = 0; ch < avctx->channels; ch++) {
143 s->samples_flt[ch] = av_malloc(avctx->frame_size *
144 sizeof(*s->samples_flt[ch]));
145 if (!s->samples_flt[ch]) {
146 ret = AVERROR(ENOMEM);
147 goto error;
148 }
149 }
150 }
151
152 ret = realloc_buffer(s);
153 if (ret < 0)
154 goto error;
155
156 avpriv_float_dsp_init(&s->fdsp, avctx->flags & CODEC_FLAG_BITEXACT);
157
158 return 0;
159 error:
160 mp3lame_encode_close(avctx);
161 return ret;
162 }
163
164 #define ENCODE_BUFFER(func, buf_type, buf_name) do { \
165 lame_result = func(s->gfp, \
166 (const buf_type *)buf_name[0], \
167 (const buf_type *)buf_name[1], frame->nb_samples, \
168 s->buffer + s->buffer_index, \
169 s->buffer_size - s->buffer_index); \
170 } while (0)
171
172 static int mp3lame_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
173 const AVFrame *frame, int *got_packet_ptr)
174 {
175 LAMEContext *s = avctx->priv_data;
176 MPADecodeHeader hdr;
177 int len, ret, ch;
178 int lame_result;
179
180 if (frame) {
181 switch (avctx->sample_fmt) {
182 case AV_SAMPLE_FMT_S16P:
183 ENCODE_BUFFER(lame_encode_buffer, int16_t, frame->data);
184 break;
185 case AV_SAMPLE_FMT_S32P:
186 ENCODE_BUFFER(lame_encode_buffer_int, int32_t, frame->data);
187 break;
188 case AV_SAMPLE_FMT_FLTP:
189 if (frame->linesize[0] < 4 * FFALIGN(frame->nb_samples, 8)) {
190 av_log(avctx, AV_LOG_ERROR, "inadequate AVFrame plane padding\n");
191 return AVERROR(EINVAL);
192 }
193 for (ch = 0; ch < avctx->channels; ch++) {
194 s->fdsp.vector_fmul_scalar(s->samples_flt[ch],
195 (const float *)frame->data[ch],
196 32768.0f,
197 FFALIGN(frame->nb_samples, 8));
198 }
199 ENCODE_BUFFER(lame_encode_buffer_float, float, s->samples_flt);
200 break;
201 default:
202 return AVERROR_BUG;
203 }
204 } else {
205 lame_result = lame_encode_flush(s->gfp, s->buffer + s->buffer_index,
206 s->buffer_size - s->buffer_index);
207 }
208 if (lame_result < 0) {
209 if (lame_result == -1) {
210 av_log(avctx, AV_LOG_ERROR,
211 "lame: output buffer too small (buffer index: %d, free bytes: %d)\n",
212 s->buffer_index, s->buffer_size - s->buffer_index);
213 }
214 return -1;
215 }
216 s->buffer_index += lame_result;
217 ret = realloc_buffer(s);
218 if (ret < 0) {
219 av_log(avctx, AV_LOG_ERROR, "error reallocating output buffer\n");
220 return ret;
221 }
222
223 /* add current frame to the queue */
224 if (frame) {
225 if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
226 return ret;
227 }
228
229 /* Move 1 frame from the LAME buffer to the output packet, if available.
230 We have to parse the first frame header in the output buffer to
231 determine the frame size. */
232 if (s->buffer_index < 4)
233 return 0;
234 if (avpriv_mpegaudio_decode_header(&hdr, AV_RB32(s->buffer))) {
235 av_log(avctx, AV_LOG_ERROR, "free format output not supported\n");
236 return -1;
237 }
238 len = hdr.frame_size;
239 av_dlog(avctx, "in:%d packet-len:%d index:%d\n", avctx->frame_size, len,
240 s->buffer_index);
241 if (len <= s->buffer_index) {
242 if ((ret = ff_alloc_packet(avpkt, len))) {
243 av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
244 return ret;
245 }
246 memcpy(avpkt->data, s->buffer, len);
247 s->buffer_index -= len;
248 memmove(s->buffer, s->buffer + len, s->buffer_index);
249
250 /* Get the next frame pts/duration */
251 ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
252 &avpkt->duration);
253
254 avpkt->size = len;
255 *got_packet_ptr = 1;
256 }
257 return 0;
258 }
259
260 #define OFFSET(x) offsetof(LAMEContext, x)
261 #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
262 static const AVOption options[] = {
263 { "reservoir", "Use bit reservoir.", OFFSET(reservoir), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, AE },
264 { "joint_stereo", "Use joint stereo.", OFFSET(joint_stereo), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, AE },
265 { NULL },
266 };
267
268 static const AVClass libmp3lame_class = {
269 .class_name = "libmp3lame encoder",
270 .item_name = av_default_item_name,
271 .option = options,
272 .version = LIBAVUTIL_VERSION_INT,
273 };
274
275 static const AVCodecDefault libmp3lame_defaults[] = {
276 { "b", "0" },
277 { NULL },
278 };
279
280 static const int libmp3lame_sample_rates[] = {
281 44100, 48000, 32000, 22050, 24000, 16000, 11025, 12000, 8000, 0
282 };
283
284 AVCodec ff_libmp3lame_encoder = {
285 .name = "libmp3lame",
286 .long_name = NULL_IF_CONFIG_SMALL("libmp3lame MP3 (MPEG audio layer 3)"),
287 .type = AVMEDIA_TYPE_AUDIO,
288 .id = AV_CODEC_ID_MP3,
289 .priv_data_size = sizeof(LAMEContext),
290 .init = mp3lame_encode_init,
291 .encode2 = mp3lame_encode_frame,
292 .close = mp3lame_encode_close,
293 .capabilities = CODEC_CAP_DELAY | CODEC_CAP_SMALL_LAST_FRAME,
294 .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S32P,
295 AV_SAMPLE_FMT_FLTP,
296 AV_SAMPLE_FMT_S16P,
297 AV_SAMPLE_FMT_NONE },
298 .supported_samplerates = libmp3lame_sample_rates,
299 .channel_layouts = (const uint64_t[]) { AV_CH_LAYOUT_MONO,
300 AV_CH_LAYOUT_STEREO,
301 0 },
302 .priv_class = &libmp3lame_class,
303 .defaults = libmp3lame_defaults,
304 };