1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-11-21 10:55:51 +02:00
FFmpeg/libavcodec/libvo-aacenc.c
Anton Khirnov 2df0c32ea1 lavc: use a separate field for exporting audio encoder padding
Currently, the amount of padding inserted at the beginning by some audio
encoders, is exported through AVCodecContext.delay. However
- the term 'delay' is heavily overloaded and can have multiple different
  meanings even in the case of audio encoding.
- this field has entirely different meanings, depending on whether the
  codec context is used for encoding or decoding (and has yet another
  different meaning for video), preventing generic handling of the codec
  context.

Therefore, add a new field -- AVCodecContext.initial_padding. It could
conceivably be used for decoding as well at a later point.
2014-10-13 19:09:01 +00:00

195 lines
6.1 KiB
C

/*
* AAC encoder wrapper
* Copyright (c) 2010 Martin Storsjo
*
* This file is part of Libav.
*
* Libav is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* Libav is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include <vo-aacenc/voAAC.h>
#include <vo-aacenc/cmnMemory.h>
#include "avcodec.h"
#include "audio_frame_queue.h"
#include "internal.h"
#include "mpeg4audio.h"
#define FRAME_SIZE 1024
#define ENC_DELAY 1600
typedef struct AACContext {
VO_AUDIO_CODECAPI codec_api;
VO_HANDLE handle;
VO_MEM_OPERATOR mem_operator;
VO_CODEC_INIT_USERDATA user_data;
VO_PBYTE end_buffer;
AudioFrameQueue afq;
int last_frame;
int last_samples;
} AACContext;
static int aac_encode_close(AVCodecContext *avctx)
{
AACContext *s = avctx->priv_data;
s->codec_api.Uninit(s->handle);
av_freep(&avctx->extradata);
ff_af_queue_close(&s->afq);
av_freep(&s->end_buffer);
return 0;
}
static av_cold int aac_encode_init(AVCodecContext *avctx)
{
AACContext *s = avctx->priv_data;
AACENC_PARAM params = { 0 };
int index, ret;
avctx->frame_size = FRAME_SIZE;
avctx->initial_padding = ENC_DELAY;
s->last_frame = 2;
ff_af_queue_init(avctx, &s->afq);
s->end_buffer = av_mallocz(avctx->frame_size * avctx->channels * 2);
if (!s->end_buffer) {
ret = AVERROR(ENOMEM);
goto error;
}
voGetAACEncAPI(&s->codec_api);
s->mem_operator.Alloc = cmnMemAlloc;
s->mem_operator.Copy = cmnMemCopy;
s->mem_operator.Free = cmnMemFree;
s->mem_operator.Set = cmnMemSet;
s->mem_operator.Check = cmnMemCheck;
s->user_data.memflag = VO_IMF_USERMEMOPERATOR;
s->user_data.memData = &s->mem_operator;
s->codec_api.Init(&s->handle, VO_AUDIO_CodingAAC, &s->user_data);
params.sampleRate = avctx->sample_rate;
params.bitRate = avctx->bit_rate;
params.nChannels = avctx->channels;
params.adtsUsed = !(avctx->flags & CODEC_FLAG_GLOBAL_HEADER);
if (s->codec_api.SetParam(s->handle, VO_PID_AAC_ENCPARAM, &params)
!= VO_ERR_NONE) {
av_log(avctx, AV_LOG_ERROR, "Unable to set encoding parameters\n");
ret = AVERROR(EINVAL);
goto error;
}
for (index = 0; index < 16; index++)
if (avctx->sample_rate == avpriv_mpeg4audio_sample_rates[index])
break;
if (index == 16) {
av_log(avctx, AV_LOG_ERROR, "Unsupported sample rate %d\n",
avctx->sample_rate);
ret = AVERROR(ENOSYS);
goto error;
}
if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
avctx->extradata_size = 2;
avctx->extradata = av_mallocz(avctx->extradata_size +
FF_INPUT_BUFFER_PADDING_SIZE);
if (!avctx->extradata) {
ret = AVERROR(ENOMEM);
goto error;
}
avctx->extradata[0] = 0x02 << 3 | index >> 1;
avctx->extradata[1] = (index & 0x01) << 7 | avctx->channels << 3;
}
return 0;
error:
aac_encode_close(avctx);
return ret;
}
static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
const AVFrame *frame, int *got_packet_ptr)
{
AACContext *s = avctx->priv_data;
VO_CODECBUFFER input = { 0 }, output = { 0 };
VO_AUDIO_OUTPUTINFO output_info = { { 0 } };
VO_PBYTE samples;
int ret;
/* handle end-of-stream small frame and flushing */
if (!frame) {
if (s->last_frame <= 0)
return 0;
if (s->last_samples > 0 && s->last_samples < ENC_DELAY - FRAME_SIZE) {
s->last_samples = 0;
s->last_frame--;
}
s->last_frame--;
memset(s->end_buffer, 0, 2 * avctx->channels * avctx->frame_size);
samples = s->end_buffer;
} else {
if (frame->nb_samples < avctx->frame_size) {
s->last_samples = frame->nb_samples;
memcpy(s->end_buffer, frame->data[0], 2 * avctx->channels * frame->nb_samples);
samples = s->end_buffer;
} else {
samples = (VO_PBYTE)frame->data[0];
}
/* add current frame to the queue */
if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
return ret;
}
if ((ret = ff_alloc_packet(avpkt, FFMAX(8192, 768 * avctx->channels)))) {
av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
return ret;
}
input.Buffer = samples;
input.Length = 2 * avctx->channels * avctx->frame_size;
output.Buffer = avpkt->data;
output.Length = avpkt->size;
s->codec_api.SetInputData(s->handle, &input);
if (s->codec_api.GetOutputData(s->handle, &output, &output_info)
!= VO_ERR_NONE) {
av_log(avctx, AV_LOG_ERROR, "Unable to encode frame\n");
return AVERROR(EINVAL);
}
/* Get the next frame pts/duration */
ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
&avpkt->duration);
avpkt->size = output.Length;
*got_packet_ptr = 1;
return 0;
}
AVCodec ff_libvo_aacenc_encoder = {
.name = "libvo_aacenc",
.long_name = NULL_IF_CONFIG_SMALL("Android VisualOn AAC (Advanced Audio Coding)"),
.type = AVMEDIA_TYPE_AUDIO,
.id = AV_CODEC_ID_AAC,
.priv_data_size = sizeof(AACContext),
.init = aac_encode_init,
.encode2 = aac_encode_frame,
.close = aac_encode_close,
.capabilities = CODEC_CAP_SMALL_LAST_FRAME | CODEC_CAP_DELAY,
.sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16,
AV_SAMPLE_FMT_NONE },
};