libmp3lame.c 8.41 KB
Newer Older
1 2 3 4
/*
 * Interface to libmp3lame for mp3 encoding
 * Copyright (c) 2002 Lennert Buytenhek <buytenh@gnu.org>
 *
5 6 7
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
8 9
 * modify it under the terms of the GNU Lesser General Public
 * 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
 * FFmpeg is distributed in the hope that it will be useful,
13
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 15
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * 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 FFmpeg; if not, write to the Free Software
19
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20
 */
21

Michael Niedermayer's avatar
Michael Niedermayer committed
22
/**
23
 * @file
Michael Niedermayer's avatar
Michael Niedermayer committed
24 25
 * Interface to libmp3lame for mp3 encoding.
 */
26

27
#include "libavutil/intreadwrite.h"
28 29 30 31
#include "avcodec.h"
#include "mpegaudio.h"
#include <lame/lame.h>

32
#define BUFFER_SIZE (7200 + 2*MPA_FRAME_SIZE + MPA_FRAME_SIZE/4)
33
typedef struct Mp3AudioContext {
34 35 36 37
    lame_global_flags *gfp;
    int stereo;
    uint8_t buffer[BUFFER_SIZE];
    int buffer_index;
38 39 40 41
    struct {
        int *left;
        int *right;
    } s32_data;
42 43
} Mp3AudioContext;

44
static av_cold int MP3lame_encode_init(AVCodecContext *avctx)
45
{
46 47 48 49 50 51 52 53 54 55 56 57
    Mp3AudioContext *s = avctx->priv_data;

    if (avctx->channels > 2)
        return -1;

    s->stereo = avctx->channels > 1 ? 1 : 0;

    if ((s->gfp = lame_init()) == NULL)
        goto err;
    lame_set_in_samplerate(s->gfp, avctx->sample_rate);
    lame_set_out_samplerate(s->gfp, avctx->sample_rate);
    lame_set_num_channels(s->gfp, avctx->channels);
58 59 60 61 62
    if(avctx->compression_level == FF_COMPRESSION_DEFAULT) {
        lame_set_quality(s->gfp, 5);
    } else {
        lame_set_quality(s->gfp, avctx->compression_level);
    }
63
    lame_set_mode(s->gfp, s->stereo ? JOINT_STEREO : MONO);
64
    lame_set_brate(s->gfp, avctx->bit_rate/1000);
65 66 67
    if(avctx->flags & CODEC_FLAG_QSCALE) {
        lame_set_brate(s->gfp, 0);
        lame_set_VBR(s->gfp, vbr_default);
68
        lame_set_VBR_quality(s->gfp, avctx->global_quality/(float)FF_QP2LAMBDA);
69
    }
70
    lame_set_bWriteVbrTag(s->gfp,0);
71
    lame_set_disable_reservoir(s->gfp, avctx->flags2 & CODEC_FLAG2_BIT_RESERVOIR ? 0 : 1);
72 73
    if (lame_init_params(s->gfp) < 0)
        goto err_close;
74

75
    avctx->frame_size = lame_get_framesize(s->gfp);
76

77 78 79 80 81
    if(!(avctx->coded_frame= avcodec_alloc_frame())) {
        lame_close(s->gfp);

        return AVERROR(ENOMEM);
    }
82
    avctx->coded_frame->key_frame= 1;
83

84 85 86 87 88 89 90 91 92 93 94 95 96
    if(AV_SAMPLE_FMT_S32 == avctx->sample_fmt && s->stereo) {
        int nelem = 2 * avctx->frame_size;

        if(! (s->s32_data.left = av_malloc(nelem * sizeof(int)))) {
            av_freep(&avctx->coded_frame);
            lame_close(s->gfp);

            return AVERROR(ENOMEM);
        }

        s->s32_data.right = s->s32_data.left + avctx->frame_size;
    }

97
    return 0;
98 99

err_close:
100
    lame_close(s->gfp);
101
err:
102
    return -1;
103 104
}

105 106
static const int sSampleRates[] = {
    44100, 48000,  32000, 22050, 24000, 16000, 11025, 12000, 8000, 0
107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133
};

static const int sBitRates[2][3][15] = {
    {   {  0, 32, 64, 96,128,160,192,224,256,288,320,352,384,416,448},
        {  0, 32, 48, 56, 64, 80, 96,112,128,160,192,224,256,320,384},
        {  0, 32, 40, 48, 56, 64, 80, 96,112,128,160,192,224,256,320}
    },
    {   {  0, 32, 48, 56, 64, 80, 96,112,128,144,160,176,192,224,256},
        {  0,  8, 16, 24, 32, 40, 48, 56, 64, 80, 96,112,128,144,160},
        {  0,  8, 16, 24, 32, 40, 48, 56, 64, 80, 96,112,128,144,160}
    },
};

static const int sSamplesPerFrame[2][3] =
{
    {  384,     1152,    1152 },
    {  384,     1152,     576 }
};

static const int sBitsPerSlot[3] = {
    32,
    8,
    8
};

static int mp3len(void *data, int *samplesPerFrame, int *sampleRate)
{
134
    uint32_t header = AV_RB32(data);
135 136 137 138 139 140 141 142 143 144 145 146 147
    int layerID = 3 - ((header >> 17) & 0x03);
    int bitRateID = ((header >> 12) & 0x0f);
    int sampleRateID = ((header >> 10) & 0x03);
    int bitsPerSlot = sBitsPerSlot[layerID];
    int isPadded = ((header >> 9) & 0x01);
    static int const mode_tab[4]= {2,3,1,0};
    int mode= mode_tab[(header >> 19) & 0x03];
    int mpeg_id= mode>0;
    int temp0, temp1, bitRate;

    if ( (( header >> 21 ) & 0x7ff) != 0x7ff || mode == 3 || layerID==3 || sampleRateID==3) {
        return -1;
    }
148

149 150 151 152 153 154 155 156 157
    if(!samplesPerFrame) samplesPerFrame= &temp0;
    if(!sampleRate     ) sampleRate     = &temp1;

//    *isMono = ((header >>  6) & 0x03) == 0x03;

    *sampleRate = sSampleRates[sampleRateID]>>mode;
    bitRate = sBitRates[mpeg_id][layerID][bitRateID] * 1000;
    *samplesPerFrame = sSamplesPerFrame[mpeg_id][layerID];
//av_log(NULL, AV_LOG_DEBUG, "sr:%d br:%d spf:%d l:%d m:%d\n", *sampleRate, bitRate, *samplesPerFrame, layerID, mode);
158

159 160 161
    return *samplesPerFrame * bitRate / (bitsPerSlot * *sampleRate) + isPadded;
}

162 163
static int MP3lame_encode_frame(AVCodecContext *avctx,
                                unsigned char *frame, int buf_size, void *data)
164
{
165 166 167
    Mp3AudioContext *s = avctx->priv_data;
    int len;
    int lame_result;
168

169
    /* lame 3.91 dies on '1-channel interleaved' data */
170

171 172 173 174 175 176 177 178
    if(!data){
        lame_result= lame_encode_flush(
                s->gfp,
                s->buffer + s->buffer_index,
                BUFFER_SIZE - s->buffer_index
                );
#if 2147483647 == INT_MAX
    }else if(AV_SAMPLE_FMT_S32 == avctx->sample_fmt){
179
        if (s->stereo) {
180 181 182 183 184 185 186 187 188 189 190
            int32_t *rp = data;
            int32_t *mp = rp + 2*avctx->frame_size;
            int *wpl = s->s32_data.left;
            int *wpr = s->s32_data.right;

            while (rp < mp) {
                *wpl++ = *rp++;
                *wpr++ = *rp++;
            }

            lame_result = lame_encode_buffer_int(
191
                s->gfp,
192 193
                s->s32_data.left,
                s->s32_data.right,
194 195
                avctx->frame_size,
                s->buffer + s->buffer_index,
196 197
                BUFFER_SIZE - s->buffer_index
                );
198
        } else {
199
            lame_result = lame_encode_buffer_int(
200 201 202
                s->gfp,
                data,
                data,
203
                avctx->frame_size,
204
                s->buffer + s->buffer_index,
205 206
                BUFFER_SIZE - s->buffer_index
                );
207
        }
208
#endif
209
    }else{
210 211
        if (s->stereo) {
            lame_result = lame_encode_buffer_interleaved(
212
                s->gfp,
213 214
                data,
                avctx->frame_size,
215
                s->buffer + s->buffer_index,
216 217
                BUFFER_SIZE - s->buffer_index
                );
218 219 220 221 222 223 224 225 226 227
        } else {
            lame_result = lame_encode_buffer(
                s->gfp,
                data,
                data,
                avctx->frame_size,
                s->buffer + s->buffer_index,
                BUFFER_SIZE - s->buffer_index
                );
        }
228 229
    }

230
    if(lame_result < 0){
Michael Niedermayer's avatar
Michael Niedermayer committed
231 232 233 234
        if(lame_result==-1) {
            /* output buffer too small */
            av_log(avctx, AV_LOG_ERROR, "lame: output buffer too small (buffer index: %d, free bytes: %d)\n", s->buffer_index, BUFFER_SIZE - s->buffer_index);
        }
235
        return -1;
236 237 238 239 240 241
    }

    s->buffer_index += lame_result;

    if(s->buffer_index<4)
        return 0;
242

Michael Niedermayer's avatar
Michael Niedermayer committed
243
    len= mp3len(s->buffer, NULL, NULL);
244
//av_log(avctx, AV_LOG_DEBUG, "in:%d packet-len:%d index:%d\n", avctx->frame_size, len, s->buffer_index);
Michael Niedermayer's avatar
Michael Niedermayer committed
245 246 247
    if(len <= s->buffer_index){
        memcpy(frame, s->buffer, len);
        s->buffer_index -= len;
248

Michael Niedermayer's avatar
Michael Niedermayer committed
249
        memmove(s->buffer, s->buffer+len, s->buffer_index);
Diego Biurrun's avatar
Diego Biurrun committed
250
            //FIXME fix the audio codec API, so we do not need the memcpy()
251
/*for(i=0; i<len; i++){
252 253
    av_log(avctx, AV_LOG_DEBUG, "%2X ", frame[i]);
}*/
Michael Niedermayer's avatar
Michael Niedermayer committed
254 255 256
        return len;
    }else
        return 0;
257 258
}

259
static av_cold int MP3lame_encode_close(AVCodecContext *avctx)
260
{
261
    Mp3AudioContext *s = avctx->priv_data;
262

263
    av_freep(&s->s32_data.left);
264
    av_freep(&avctx->coded_frame);
265

266 267
    lame_close(s->gfp);
    return 0;
268 269 270
}


271
AVCodec ff_libmp3lame_encoder = {
272
    "libmp3lame",
273
    AVMEDIA_TYPE_AUDIO,
274
    CODEC_ID_MP3,
275 276 277
    sizeof(Mp3AudioContext),
    MP3lame_encode_init,
    MP3lame_encode_frame,
278 279
    MP3lame_encode_close,
    .capabilities= CODEC_CAP_DELAY,
280 281 282 283 284
    .sample_fmts = (const enum AVSampleFormat[]){AV_SAMPLE_FMT_S16,
#if 2147483647 == INT_MAX
    AV_SAMPLE_FMT_S32,
#endif
    AV_SAMPLE_FMT_NONE},
285
    .supported_samplerates= sSampleRates,
286
    .long_name= NULL_IF_CONFIG_SMALL("libmp3lame MP3 (MPEG audio layer 3)"),
287
};