libavcodec/libmp3lame.c
0716b577
 /*
  * Interface to libmp3lame for mp3 encoding
  * Copyright (c) 2002 Lennert Buytenhek <buytenh@gnu.org>
  *
b78e7197
  * This file is part of FFmpeg.
  *
  * FFmpeg is free software; you can redistribute it and/or
ff4ec49e
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
b78e7197
  * version 2.1 of the License, or (at your option) any later version.
0716b577
  *
b78e7197
  * FFmpeg is distributed in the hope that it will be useful,
0716b577
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
ff4ec49e
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  * Lesser General Public License for more details.
0716b577
  *
ff4ec49e
  * You should have received a copy of the GNU Lesser General Public
b78e7197
  * License along with FFmpeg; if not, write to the Free Software
5509bffa
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
0716b577
  */
115329f1
 
983e3246
 /**
ba87f080
  * @file
983e3246
  * Interface to libmp3lame for mp3 encoding.
  */
0716b577
 
e2322252
 #include <lame/lame.h>
 
a903f8f0
 #include "libavutil/channel_layout.h"
1d9c2dc8
 #include "libavutil/common.h"
284ea790
 #include "libavutil/float_dsp.h"
ed96fffb
 #include "libavutil/intreadwrite.h"
a7cec3a0
 #include "libavutil/log.h"
 #include "libavutil/opt.h"
0716b577
 #include "avcodec.h"
3d853d7a
 #include "audio_frame_queue.h"
232e16dd
 #include "internal.h"
0716b577
 #include "mpegaudio.h"
1f516c04
 #include "mpegaudiodecheader.h"
0716b577
 
84d0fcf2
 #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.
e2322252
 
 typedef struct LAMEContext {
a7cec3a0
     AVClass *class;
e00959a9
     AVCodecContext *avctx;
e344c1ea
     lame_global_flags *gfp;
abd8b9e7
     uint8_t *buffer;
e344c1ea
     int buffer_index;
abd8b9e7
     int buffer_size;
a7cec3a0
     int reservoir;
1de11a7c
     int joint_stereo;
d3211cfa
     int abr;
473b297f
     float *samples_flt[2];
3d853d7a
     AudioFrameQueue afq;
284ea790
     AVFloatDSPContext fdsp;
e2322252
 } LAMEContext;
0716b577
 
 
abd8b9e7
 static int realloc_buffer(LAMEContext *s)
 {
     if (!s->buffer || s->buffer_size - s->buffer_index < BUFFER_SIZE) {
9b8d11a7
         int new_size = s->buffer_index + 2 * BUFFER_SIZE, err;
abd8b9e7
 
         av_dlog(s->avctx, "resizing output buffer: %d -> %d\n", s->buffer_size,
                 new_size);
9b8d11a7
         if ((err = av_reallocp(&s->buffer, new_size)) < 0) {
abd8b9e7
             s->buffer_size = s->buffer_index = 0;
9b8d11a7
             return err;
abd8b9e7
         }
         s->buffer_size = new_size;
     }
     return 0;
 }
 
e2322252
 static av_cold int mp3lame_encode_close(AVCodecContext *avctx)
0716b577
 {
e2322252
     LAMEContext *s = avctx->priv_data;
e344c1ea
 
473b297f
     av_freep(&s->samples_flt[0]);
     av_freep(&s->samples_flt[1]);
abd8b9e7
     av_freep(&s->buffer);
8dad25eb
 
3d853d7a
     ff_af_queue_close(&s->afq);
 
8dad25eb
     lame_close(s->gfp);
     return 0;
 }
 
e2322252
 static av_cold int mp3lame_encode_init(AVCodecContext *avctx)
0716b577
 {
e2322252
     LAMEContext *s = avctx->priv_data;
8dad25eb
     int ret;
e344c1ea
 
e00959a9
     s->avctx = avctx;
 
e2322252
     /* initialize LAME and get defaults */
e344c1ea
     if ((s->gfp = lame_init()) == NULL)
8dad25eb
         return AVERROR(ENOMEM);
e2322252
 
1cbf7fb4
 
e344c1ea
     lame_set_num_channels(s->gfp, avctx->channels);
1de11a7c
     lame_set_mode(s->gfp, avctx->channels > 1 ? s->joint_stereo ? JOINT_STEREO : STEREO : MONO);
e344c1ea
 
e2322252
     /* sample rate */
     lame_set_in_samplerate (s->gfp, avctx->sample_rate);
e344c1ea
     lame_set_out_samplerate(s->gfp, avctx->sample_rate);
e2322252
 
     /* algorithmic quality */
     if (avctx->compression_level == FF_COMPRESSION_DEFAULT)
afffd357
         lame_set_quality(s->gfp, 5);
e2322252
     else
afffd357
         lame_set_quality(s->gfp, avctx->compression_level);
e2322252
 
     /* rate control */
07e7bc9c
     if (avctx->flags & CODEC_FLAG_QSCALE) { // VBR
c57c770d
         lame_set_VBR(s->gfp, vbr_default);
c4db3446
         lame_set_VBR_quality(s->gfp, avctx->global_quality / (float)FF_QP2LAMBDA);
232e16dd
     } else {
d3211cfa
         if (avctx->bit_rate) {
             if (s->abr) {                   // ABR
                 lame_set_VBR(s->gfp, vbr_abr);
                 lame_set_VBR_mean_bitrate_kbps(s->gfp, avctx->bit_rate / 1000);
             } else                          // CBR
                 lame_set_brate(s->gfp, avctx->bit_rate / 1000);
         }
c57c770d
     }
0716b577
 
e2322252
     /* do not get a Xing VBR header frame from LAME */
e344c1ea
     lame_set_bWriteVbrTag(s->gfp,0);
115329f1
 
e2322252
     /* bit reservoir usage */
a7cec3a0
     lame_set_disable_reservoir(s->gfp, !s->reservoir);
22fa406f
 
e2322252
     /* set specified parameters */
8dad25eb
     if (lame_init_params(s->gfp) < 0) {
         ret = -1;
         goto error;
22fa406f
     }
0716b577
 
3d853d7a
     /* get encoder delay */
     avctx->delay = lame_get_encoder_delay(s->gfp) + 528 + 1;
     ff_af_queue_init(avctx, &s->afq);
 
e2322252
     avctx->frame_size  = lame_get_framesize(s->gfp);
3d853d7a
 
473b297f
     /* allocate float sample buffers */
     if (avctx->sample_fmt == AV_SAMPLE_FMT_FLTP) {
e00959a9
         int ch;
         for (ch = 0; ch < avctx->channels; ch++) {
473b297f
             s->samples_flt[ch] = av_malloc(avctx->frame_size *
                                            sizeof(*s->samples_flt[ch]));
             if (!s->samples_flt[ch]) {
e00959a9
                 ret = AVERROR(ENOMEM);
                 goto error;
             }
39e69d0b
         }
     }
 
abd8b9e7
     ret = realloc_buffer(s);
     if (ret < 0)
         goto error;
 
284ea790
     avpriv_float_dsp_init(&s->fdsp, avctx->flags & CODEC_FLAG_BITEXACT);
473b297f
 
e344c1ea
     return 0;
8dad25eb
 error:
e2322252
     mp3lame_encode_close(avctx);
8dad25eb
     return ret;
0716b577
 }
 
473b297f
 #define ENCODE_BUFFER(func, buf_type, buf_name) do {                        \
     lame_result = func(s->gfp,                                              \
                        (const buf_type *)buf_name[0],                       \
                        (const buf_type *)buf_name[1], frame->nb_samples,    \
                        s->buffer + s->buffer_index,                         \
abd8b9e7
                        s->buffer_size - s->buffer_index);                   \
e00959a9
 } while (0)
 
3d853d7a
 static int mp3lame_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
                                 const AVFrame *frame, int *got_packet_ptr)
0716b577
 {
e2322252
     LAMEContext *s = avctx->priv_data;
1f516c04
     MPADecodeHeader hdr;
473b297f
     int len, ret, ch;
e344c1ea
     int lame_result;
0716b577
 
3d853d7a
     if (frame) {
e00959a9
         switch (avctx->sample_fmt) {
473b297f
         case AV_SAMPLE_FMT_S16P:
             ENCODE_BUFFER(lame_encode_buffer, int16_t, frame->data);
e00959a9
             break;
473b297f
         case AV_SAMPLE_FMT_S32P:
             ENCODE_BUFFER(lame_encode_buffer_int, int32_t, frame->data);
e00959a9
             break;
473b297f
         case AV_SAMPLE_FMT_FLTP:
             if (frame->linesize[0] < 4 * FFALIGN(frame->nb_samples, 8)) {
                 av_log(avctx, AV_LOG_ERROR, "inadequate AVFrame plane padding\n");
                 return AVERROR(EINVAL);
             }
             for (ch = 0; ch < avctx->channels; ch++) {
284ea790
                 s->fdsp.vector_fmul_scalar(s->samples_flt[ch],
                                            (const float *)frame->data[ch],
                                            32768.0f,
                                            FFALIGN(frame->nb_samples, 8));
473b297f
             }
             ENCODE_BUFFER(lame_encode_buffer_float, float, s->samples_flt);
e00959a9
             break;
         default:
             return AVERROR_BUG;
39e69d0b
         }
c4db3446
     } else {
         lame_result = lame_encode_flush(s->gfp, s->buffer + s->buffer_index,
e984f478
                                         s->buffer_size - s->buffer_index);
0f5c3f21
     }
c4db3446
     if (lame_result < 0) {
         if (lame_result == -1) {
             av_log(avctx, AV_LOG_ERROR,
                    "lame: output buffer too small (buffer index: %d, free bytes: %d)\n",
abd8b9e7
                    s->buffer_index, s->buffer_size - s->buffer_index);
203fa6b4
         }
20836483
         return -1;
0f5c3f21
     }
     s->buffer_index += lame_result;
abd8b9e7
     ret = realloc_buffer(s);
     if (ret < 0) {
         av_log(avctx, AV_LOG_ERROR, "error reallocating output buffer\n");
         return ret;
     }
0f5c3f21
 
3d853d7a
     /* add current frame to the queue */
     if (frame) {
871b6ec0
         if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
3d853d7a
             return ret;
     }
 
e2322252
     /* Move 1 frame from the LAME buffer to the output packet, if available.
        We have to parse the first frame header in the output buffer to
        determine the frame size. */
c4db3446
     if (s->buffer_index < 4)
0f5c3f21
         return 0;
1f516c04
     if (avpriv_mpegaudio_decode_header(&hdr, AV_RB32(s->buffer))) {
         av_log(avctx, AV_LOG_ERROR, "free format output not supported\n");
         return -1;
     }
     len = hdr.frame_size;
bf909fc4
     av_dlog(avctx, "in:%d packet-len:%d index:%d\n", avctx->frame_size, len,
             s->buffer_index);
c4db3446
     if (len <= s->buffer_index) {
bcaf64b6
         if ((ret = ff_alloc_packet2(avctx, avpkt, len)) < 0)
3d853d7a
             return ret;
         memcpy(avpkt->data, s->buffer, len);
203fa6b4
         s->buffer_index -= len;
c4db3446
         memmove(s->buffer, s->buffer + len, s->buffer_index);
3d853d7a
 
         /* Get the next frame pts/duration */
         ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
                            &avpkt->duration);
 
         avpkt->size = len;
         *got_packet_ptr = 1;
     }
     return 0;
0716b577
 }
 
e2322252
 #define OFFSET(x) offsetof(LAMEContext, x)
a7cec3a0
 #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
 static const AVOption options[] = {
683d9cb1
     { "reservoir",    "use bit reservoir", OFFSET(reservoir),    AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, AE },
     { "joint_stereo", "use joint stereo",  OFFSET(joint_stereo), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, AE },
d3211cfa
     { "abr",          "use ABR",           OFFSET(abr),          AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AE },
a7cec3a0
     { NULL },
 };
 
 static const AVClass libmp3lame_class = {
     .class_name = "libmp3lame encoder",
     .item_name  = av_default_item_name,
     .option     = options,
     .version    = LIBAVUTIL_VERSION_INT,
 };
0716b577
 
232e16dd
 static const AVCodecDefault libmp3lame_defaults[] = {
     { "b",          "0" },
     { NULL },
 };
 
e2322252
 static const int libmp3lame_sample_rates[] = {
     44100, 48000,  32000, 22050, 24000, 16000, 11025, 12000, 8000, 0
 };
 
e7e2df27
 AVCodec ff_libmp3lame_encoder = {
c4db3446
     .name                  = "libmp3lame",
b2bed932
     .long_name             = NULL_IF_CONFIG_SMALL("libmp3lame MP3 (MPEG audio layer 3)"),
c4db3446
     .type                  = AVMEDIA_TYPE_AUDIO,
36ef5369
     .id                    = AV_CODEC_ID_MP3,
e2322252
     .priv_data_size        = sizeof(LAMEContext),
     .init                  = mp3lame_encode_init,
3d853d7a
     .encode2               = mp3lame_encode_frame,
e2322252
     .close                 = mp3lame_encode_close,
3d853d7a
     .capabilities          = CODEC_CAP_DELAY | CODEC_CAP_SMALL_LAST_FRAME,
473b297f
     .sample_fmts           = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S32P,
                                                              AV_SAMPLE_FMT_FLTP,
                                                              AV_SAMPLE_FMT_S16P,
c4db3446
                                                              AV_SAMPLE_FMT_NONE },
e2322252
     .supported_samplerates = libmp3lame_sample_rates,
e6694659
     .channel_layouts       = (const uint64_t[]) { AV_CH_LAYOUT_MONO,
e47e2369
                                                   AV_CH_LAYOUT_STEREO,
bcbb30e2
                                                   0 },
c4db3446
     .priv_class            = &libmp3lame_class,
232e16dd
     .defaults              = libmp3lame_defaults,
0716b577
 };