| 1 | /* |
| 2 | * Interface to libmp3lame for mp3 encoding |
| 3 | * Copyright (c) 2002 Lennert Buytenhek <buytenh@gnu.org> |
| 4 | * |
| 5 | * This file is part of FFmpeg. |
| 6 | * |
| 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. |
| 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 | * 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 FFmpeg; 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+1000) // FIXME: Buffer size to small? Adding 1000 to make up for it. |
| 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 | int abr; |
| 53 | float *samples_flt[2]; |
| 54 | AudioFrameQueue afq; |
| 55 | AVFloatDSPContext *fdsp; |
| 56 | } LAMEContext; |
| 57 | |
| 58 | |
| 59 | static int realloc_buffer(LAMEContext *s) |
| 60 | { |
| 61 | if (!s->buffer || s->buffer_size - s->buffer_index < BUFFER_SIZE) { |
| 62 | int new_size = s->buffer_index + 2 * BUFFER_SIZE, err; |
| 63 | |
| 64 | av_dlog(s->avctx, "resizing output buffer: %d -> %d\n", s->buffer_size, |
| 65 | new_size); |
| 66 | if ((err = av_reallocp(&s->buffer, new_size)) < 0) { |
| 67 | s->buffer_size = s->buffer_index = 0; |
| 68 | return err; |
| 69 | } |
| 70 | s->buffer_size = new_size; |
| 71 | } |
| 72 | return 0; |
| 73 | } |
| 74 | |
| 75 | static av_cold int mp3lame_encode_close(AVCodecContext *avctx) |
| 76 | { |
| 77 | LAMEContext *s = avctx->priv_data; |
| 78 | |
| 79 | av_freep(&s->samples_flt[0]); |
| 80 | av_freep(&s->samples_flt[1]); |
| 81 | av_freep(&s->buffer); |
| 82 | av_freep(&s->fdsp); |
| 83 | |
| 84 | ff_af_queue_close(&s->afq); |
| 85 | |
| 86 | lame_close(s->gfp); |
| 87 | return 0; |
| 88 | } |
| 89 | |
| 90 | static av_cold int mp3lame_encode_init(AVCodecContext *avctx) |
| 91 | { |
| 92 | LAMEContext *s = avctx->priv_data; |
| 93 | int ret; |
| 94 | |
| 95 | s->avctx = avctx; |
| 96 | |
| 97 | /* initialize LAME and get defaults */ |
| 98 | if (!(s->gfp = lame_init())) |
| 99 | return AVERROR(ENOMEM); |
| 100 | |
| 101 | |
| 102 | lame_set_num_channels(s->gfp, avctx->channels); |
| 103 | lame_set_mode(s->gfp, avctx->channels > 1 ? s->joint_stereo ? JOINT_STEREO : STEREO : MONO); |
| 104 | |
| 105 | /* sample rate */ |
| 106 | lame_set_in_samplerate (s->gfp, avctx->sample_rate); |
| 107 | lame_set_out_samplerate(s->gfp, avctx->sample_rate); |
| 108 | |
| 109 | /* algorithmic quality */ |
| 110 | if (avctx->compression_level != FF_COMPRESSION_DEFAULT) |
| 111 | lame_set_quality(s->gfp, avctx->compression_level); |
| 112 | |
| 113 | /* rate control */ |
| 114 | if (avctx->flags & CODEC_FLAG_QSCALE) { // VBR |
| 115 | lame_set_VBR(s->gfp, vbr_default); |
| 116 | lame_set_VBR_quality(s->gfp, avctx->global_quality / (float)FF_QP2LAMBDA); |
| 117 | } else { |
| 118 | if (avctx->bit_rate) { |
| 119 | if (s->abr) { // ABR |
| 120 | lame_set_VBR(s->gfp, vbr_abr); |
| 121 | lame_set_VBR_mean_bitrate_kbps(s->gfp, avctx->bit_rate / 1000); |
| 122 | } else // CBR |
| 123 | lame_set_brate(s->gfp, avctx->bit_rate / 1000); |
| 124 | } |
| 125 | } |
| 126 | |
| 127 | /* do not get a Xing VBR header frame from LAME */ |
| 128 | lame_set_bWriteVbrTag(s->gfp,0); |
| 129 | |
| 130 | /* bit reservoir usage */ |
| 131 | lame_set_disable_reservoir(s->gfp, !s->reservoir); |
| 132 | |
| 133 | /* set specified parameters */ |
| 134 | if (lame_init_params(s->gfp) < 0) { |
| 135 | ret = -1; |
| 136 | goto error; |
| 137 | } |
| 138 | |
| 139 | /* get encoder delay */ |
| 140 | avctx->initial_padding = lame_get_encoder_delay(s->gfp) + 528 + 1; |
| 141 | ff_af_queue_init(avctx, &s->afq); |
| 142 | |
| 143 | avctx->frame_size = lame_get_framesize(s->gfp); |
| 144 | |
| 145 | /* allocate float sample buffers */ |
| 146 | if (avctx->sample_fmt == AV_SAMPLE_FMT_FLTP) { |
| 147 | int ch; |
| 148 | for (ch = 0; ch < avctx->channels; ch++) { |
| 149 | s->samples_flt[ch] = av_malloc(avctx->frame_size * |
| 150 | sizeof(*s->samples_flt[ch])); |
| 151 | if (!s->samples_flt[ch]) { |
| 152 | ret = AVERROR(ENOMEM); |
| 153 | goto error; |
| 154 | } |
| 155 | } |
| 156 | } |
| 157 | |
| 158 | ret = realloc_buffer(s); |
| 159 | if (ret < 0) |
| 160 | goto error; |
| 161 | |
| 162 | s->fdsp = avpriv_float_dsp_alloc(avctx->flags & CODEC_FLAG_BITEXACT); |
| 163 | if (!s->fdsp) { |
| 164 | ret = AVERROR(ENOMEM); |
| 165 | goto error; |
| 166 | } |
| 167 | |
| 168 | |
| 169 | return 0; |
| 170 | error: |
| 171 | mp3lame_encode_close(avctx); |
| 172 | return ret; |
| 173 | } |
| 174 | |
| 175 | #define ENCODE_BUFFER(func, buf_type, buf_name) do { \ |
| 176 | lame_result = func(s->gfp, \ |
| 177 | (const buf_type *)buf_name[0], \ |
| 178 | (const buf_type *)buf_name[1], frame->nb_samples, \ |
| 179 | s->buffer + s->buffer_index, \ |
| 180 | s->buffer_size - s->buffer_index); \ |
| 181 | } while (0) |
| 182 | |
| 183 | static int mp3lame_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, |
| 184 | const AVFrame *frame, int *got_packet_ptr) |
| 185 | { |
| 186 | LAMEContext *s = avctx->priv_data; |
| 187 | MPADecodeHeader hdr; |
| 188 | int len, ret, ch; |
| 189 | int lame_result; |
| 190 | uint32_t h; |
| 191 | |
| 192 | if (frame) { |
| 193 | switch (avctx->sample_fmt) { |
| 194 | case AV_SAMPLE_FMT_S16P: |
| 195 | ENCODE_BUFFER(lame_encode_buffer, int16_t, frame->data); |
| 196 | break; |
| 197 | case AV_SAMPLE_FMT_S32P: |
| 198 | ENCODE_BUFFER(lame_encode_buffer_int, int32_t, frame->data); |
| 199 | break; |
| 200 | case AV_SAMPLE_FMT_FLTP: |
| 201 | if (frame->linesize[0] < 4 * FFALIGN(frame->nb_samples, 8)) { |
| 202 | av_log(avctx, AV_LOG_ERROR, "inadequate AVFrame plane padding\n"); |
| 203 | return AVERROR(EINVAL); |
| 204 | } |
| 205 | for (ch = 0; ch < avctx->channels; ch++) { |
| 206 | s->fdsp->vector_fmul_scalar(s->samples_flt[ch], |
| 207 | (const float *)frame->data[ch], |
| 208 | 32768.0f, |
| 209 | FFALIGN(frame->nb_samples, 8)); |
| 210 | } |
| 211 | ENCODE_BUFFER(lame_encode_buffer_float, float, s->samples_flt); |
| 212 | break; |
| 213 | default: |
| 214 | return AVERROR_BUG; |
| 215 | } |
| 216 | } else if (!s->afq.frame_alloc) { |
| 217 | lame_result = 0; |
| 218 | } else { |
| 219 | lame_result = lame_encode_flush(s->gfp, s->buffer + s->buffer_index, |
| 220 | s->buffer_size - s->buffer_index); |
| 221 | } |
| 222 | if (lame_result < 0) { |
| 223 | if (lame_result == -1) { |
| 224 | av_log(avctx, AV_LOG_ERROR, |
| 225 | "lame: output buffer too small (buffer index: %d, free bytes: %d)\n", |
| 226 | s->buffer_index, s->buffer_size - s->buffer_index); |
| 227 | } |
| 228 | return -1; |
| 229 | } |
| 230 | s->buffer_index += lame_result; |
| 231 | ret = realloc_buffer(s); |
| 232 | if (ret < 0) { |
| 233 | av_log(avctx, AV_LOG_ERROR, "error reallocating output buffer\n"); |
| 234 | return ret; |
| 235 | } |
| 236 | |
| 237 | /* add current frame to the queue */ |
| 238 | if (frame) { |
| 239 | if ((ret = ff_af_queue_add(&s->afq, frame)) < 0) |
| 240 | return ret; |
| 241 | } |
| 242 | |
| 243 | /* Move 1 frame from the LAME buffer to the output packet, if available. |
| 244 | We have to parse the first frame header in the output buffer to |
| 245 | determine the frame size. */ |
| 246 | if (s->buffer_index < 4) |
| 247 | return 0; |
| 248 | h = AV_RB32(s->buffer); |
| 249 | if (ff_mpa_check_header(h) < 0) { |
| 250 | av_log(avctx, AV_LOG_ERROR, "Invalid mp3 header at start of buffer\n"); |
| 251 | return AVERROR_BUG; |
| 252 | } |
| 253 | if (avpriv_mpegaudio_decode_header(&hdr, h)) { |
| 254 | av_log(avctx, AV_LOG_ERROR, "free format output not supported\n"); |
| 255 | return -1; |
| 256 | } |
| 257 | len = hdr.frame_size; |
| 258 | av_dlog(avctx, "in:%d packet-len:%d index:%d\n", avctx->frame_size, len, |
| 259 | s->buffer_index); |
| 260 | if (len <= s->buffer_index) { |
| 261 | if ((ret = ff_alloc_packet2(avctx, avpkt, len)) < 0) |
| 262 | return ret; |
| 263 | memcpy(avpkt->data, s->buffer, len); |
| 264 | s->buffer_index -= len; |
| 265 | memmove(s->buffer, s->buffer + len, s->buffer_index); |
| 266 | |
| 267 | /* Get the next frame pts/duration */ |
| 268 | ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts, |
| 269 | &avpkt->duration); |
| 270 | |
| 271 | avpkt->size = len; |
| 272 | *got_packet_ptr = 1; |
| 273 | } |
| 274 | return 0; |
| 275 | } |
| 276 | |
| 277 | #define OFFSET(x) offsetof(LAMEContext, x) |
| 278 | #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM |
| 279 | static const AVOption options[] = { |
| 280 | { "reservoir", "use bit reservoir", OFFSET(reservoir), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, AE }, |
| 281 | { "joint_stereo", "use joint stereo", OFFSET(joint_stereo), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, AE }, |
| 282 | { "abr", "use ABR", OFFSET(abr), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AE }, |
| 283 | { NULL }, |
| 284 | }; |
| 285 | |
| 286 | static const AVClass libmp3lame_class = { |
| 287 | .class_name = "libmp3lame encoder", |
| 288 | .item_name = av_default_item_name, |
| 289 | .option = options, |
| 290 | .version = LIBAVUTIL_VERSION_INT, |
| 291 | }; |
| 292 | |
| 293 | static const AVCodecDefault libmp3lame_defaults[] = { |
| 294 | { "b", "0" }, |
| 295 | { NULL }, |
| 296 | }; |
| 297 | |
| 298 | static const int libmp3lame_sample_rates[] = { |
| 299 | 44100, 48000, 32000, 22050, 24000, 16000, 11025, 12000, 8000, 0 |
| 300 | }; |
| 301 | |
| 302 | AVCodec ff_libmp3lame_encoder = { |
| 303 | .name = "libmp3lame", |
| 304 | .long_name = NULL_IF_CONFIG_SMALL("libmp3lame MP3 (MPEG audio layer 3)"), |
| 305 | .type = AVMEDIA_TYPE_AUDIO, |
| 306 | .id = AV_CODEC_ID_MP3, |
| 307 | .priv_data_size = sizeof(LAMEContext), |
| 308 | .init = mp3lame_encode_init, |
| 309 | .encode2 = mp3lame_encode_frame, |
| 310 | .close = mp3lame_encode_close, |
| 311 | .capabilities = CODEC_CAP_DELAY | CODEC_CAP_SMALL_LAST_FRAME, |
| 312 | .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S32P, |
| 313 | AV_SAMPLE_FMT_FLTP, |
| 314 | AV_SAMPLE_FMT_S16P, |
| 315 | AV_SAMPLE_FMT_NONE }, |
| 316 | .supported_samplerates = libmp3lame_sample_rates, |
| 317 | .channel_layouts = (const uint64_t[]) { AV_CH_LAYOUT_MONO, |
| 318 | AV_CH_LAYOUT_STEREO, |
| 319 | 0 }, |
| 320 | .priv_class = &libmp3lame_class, |
| 321 | .defaults = libmp3lame_defaults, |
| 322 | }; |