1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00

avcodec: add QOA decoder

This commit is contained in:
Paul B Mahol 2023-09-23 16:49:25 +02:00
parent 93b5d9030b
commit 3609d2b783
7 changed files with 187 additions and 1 deletions

View File

@ -5,6 +5,7 @@ version <next>:
- LEAD MCMP decoder
- EVC decoding using external library libxevd
- EVC encoding using external library libxeve
- QOA decoder
version 6.1:
- libaribcaption decoder

View File

@ -623,6 +623,7 @@ OBJS-$(CONFIG_QCELP_DECODER) += qcelpdec.o \
OBJS-$(CONFIG_QDM2_DECODER) += qdm2.o
OBJS-$(CONFIG_QDMC_DECODER) += qdmc.o
OBJS-$(CONFIG_QDRAW_DECODER) += qdrw.o
OBJS-$(CONFIG_QOA_DECODER) += qoadec.o
OBJS-$(CONFIG_QOI_DECODER) += qoidec.o
OBJS-$(CONFIG_QOI_ENCODER) += qoienc.o
OBJS-$(CONFIG_QPEG_DECODER) += qpeg.o

View File

@ -522,6 +522,7 @@ extern const FFCodec ff_paf_audio_decoder;
extern const FFCodec ff_qcelp_decoder;
extern const FFCodec ff_qdm2_decoder;
extern const FFCodec ff_qdmc_decoder;
extern const FFCodec ff_qoa_decoder;
extern const FFCodec ff_ra_144_encoder;
extern const FFCodec ff_ra_144_decoder;
extern const FFCodec ff_ra_288_decoder;

View File

@ -3427,6 +3427,13 @@ static const AVCodecDescriptor codec_descriptors[] = {
.long_name = NULL_IF_CONFIG_SMALL("OSQ (Original Sound Quality)"),
.props = AV_CODEC_PROP_INTRA_ONLY | AV_CODEC_PROP_LOSSLESS,
},
{
.id = AV_CODEC_ID_QOA,
.type = AVMEDIA_TYPE_AUDIO,
.name = "qoa",
.long_name = NULL_IF_CONFIG_SMALL("QOA (Quite OK Audio)"),
.props = AV_CODEC_PROP_INTRA_ONLY | AV_CODEC_PROP_LOSSY,
},
/* subtitle codecs */
{

View File

@ -545,6 +545,7 @@ enum AVCodecID {
AV_CODEC_ID_RKA,
AV_CODEC_ID_AC4,
AV_CODEC_ID_OSQ,
AV_CODEC_ID_QOA,
/* subtitle codecs */
AV_CODEC_ID_FIRST_SUBTITLE = 0x17000, ///< A dummy ID pointing at the start of subtitle codecs.

175
libavcodec/qoadec.c Normal file
View File

@ -0,0 +1,175 @@
/*
* QOA decoder
*
* This file is part of FFmpeg.
*
* FFmpeg 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.
*
* FFmpeg 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 FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "avcodec.h"
#include "codec_internal.h"
#include "decode.h"
#include "get_bits.h"
#include "bytestream.h"
#include "mathops.h"
#define QOA_SLICE_LEN 20
#define QOA_LMS_LEN 4
typedef struct QOAChannel {
int history[QOA_LMS_LEN];
int weights[QOA_LMS_LEN];
} QOAChannel;
typedef struct QOAContext {
QOAChannel *ch;
} QOAContext;
static const int16_t qoa_dequant_tab[16][8] = {
{ 1, -1, 3, -3, 5, -5, 7, -7},
{ 5, -5, 18, -18, 32, -32, 49, -49},
{ 16, -16, 53, -53, 95, -95, 147, -147},
{ 34, -34, 113, -113, 203, -203, 315, -315},
{ 63, -63, 210, -210, 378, -378, 588, -588},
{ 104, -104, 345, -345, 621, -621, 966, -966},
{ 158, -158, 528, -528, 950, -950, 1477, -1477},
{ 228, -228, 760, -760, 1368, -1368, 2128, -2128},
{ 316, -316, 1053, -1053, 1895, -1895, 2947, -2947},
{ 422, -422, 1405, -1405, 2529, -2529, 3934, -3934},
{ 548, -548, 1828, -1828, 3290, -3290, 5117, -5117},
{ 696, -696, 2320, -2320, 4176, -4176, 6496, -6496},
{ 868, -868, 2893, -2893, 5207, -5207, 8099, -8099},
{1064, -1064, 3548, -3548, 6386, -6386, 9933, -9933},
{1286, -1286, 4288, -4288, 7718, -7718, 12005, -12005},
{1536, -1536, 5120, -5120, 9216, -9216, 14336, -14336},
};
static av_cold int qoa_decode_init(AVCodecContext *avctx)
{
QOAContext *s = avctx->priv_data;
avctx->sample_fmt = AV_SAMPLE_FMT_S16;
s->ch = av_calloc(avctx->ch_layout.nb_channels, sizeof(*s->ch));
if (!s->ch)
return AVERROR(ENOMEM);
return 0;
}
static int qoa_lms_predict(QOAChannel *lms)
{
int prediction = 0;
for (int i = 0; i < QOA_LMS_LEN; i++)
prediction += lms->weights[i] * lms->history[i];
return prediction >> 13;
}
static void qoa_lms_update(QOAChannel *lms, int sample, int residual)
{
int delta = residual >> 4;
for (int i = 0; i < QOA_LMS_LEN; i++)
lms->weights[i] += lms->history[i] < 0 ? -delta : delta;
for (int i = 0; i < QOA_LMS_LEN-1; i++)
lms->history[i] = lms->history[i+1];
lms->history[QOA_LMS_LEN-1] = sample;
}
static int qoa_decode_frame(AVCodecContext *avctx, AVFrame *frame,
int *got_frame_ptr, AVPacket *avpkt)
{
QOAContext *s = avctx->priv_data;
int ret, frame_size, nb_channels;
GetByteContext gb;
int16_t *samples;
bytestream2_init(&gb, avpkt->data, avpkt->size);
nb_channels = bytestream2_get_byte(&gb);
if (avctx->ch_layout.nb_channels != nb_channels)
return AVERROR_INVALIDDATA;
avctx->sample_rate = bytestream2_get_be24(&gb);
frame->nb_samples = bytestream2_get_be16(&gb);
frame_size = bytestream2_get_be16(&gb);
if (frame_size > avpkt->size)
return AVERROR_INVALIDDATA;
if (frame_size < 8 + QOA_LMS_LEN * 4 * nb_channels +
8LL * frame->nb_samples * nb_channels / QOA_SLICE_LEN)
return AVERROR_INVALIDDATA;
if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
return ret;
samples = (int16_t *)frame->data[0];
for (int ch = 0; ch < nb_channels; ch++) {
QOAChannel *qch = &s->ch[ch];
for (int n = 0; n < QOA_LMS_LEN; n++)
qch->history[n] = sign_extend(bytestream2_get_be16u(&gb), 16);
for (int n = 0; n < QOA_LMS_LEN; n++)
qch->weights[n] = sign_extend(bytestream2_get_be16u(&gb), 16);
}
for (int sample_index = 0; sample_index < frame->nb_samples * nb_channels;
sample_index += QOA_SLICE_LEN) {
for (int ch = 0; ch < nb_channels; ch++) {
QOAChannel *lms = &s->ch[ch];
uint64_t slice = bytestream2_get_be64u(&gb);
int scalefactor = (slice >> 60) & 0xf;
int slice_start = sample_index * nb_channels + ch;
int slice_end = av_clip(sample_index + QOA_SLICE_LEN, 0, frame->nb_samples) * nb_channels + ch;
for (int si = slice_start; si < slice_end; si += nb_channels) {
int predicted = qoa_lms_predict(lms);
int quantized = (slice >> 57) & 0x7;
int dequantized = qoa_dequant_tab[scalefactor][quantized];
int reconstructed = av_clip_int16(predicted + dequantized);
samples[si] = reconstructed;
slice <<= 3;
qoa_lms_update(lms, reconstructed, dequantized);
}
}
}
*got_frame_ptr = 1;
return avpkt->size;
}
static av_cold int qoa_decode_end(AVCodecContext *avctx)
{
QOAContext *s = avctx->priv_data;
av_freep(&s->ch);
return 0;
}
const FFCodec ff_qoa_decoder = {
.p.name = "qoa",
CODEC_LONG_NAME("QOA (Quite OK Audio)"),
.p.type = AVMEDIA_TYPE_AUDIO,
.p.id = AV_CODEC_ID_QOA,
.priv_data_size = sizeof(QOAContext),
.init = qoa_decode_init,
FF_CODEC_DECODE_CB(qoa_decode_frame),
.close = qoa_decode_end,
.p.capabilities = AV_CODEC_CAP_CHANNEL_CONF |
AV_CODEC_CAP_DR1,
.p.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S16,
AV_SAMPLE_FMT_NONE },
};

View File

@ -29,7 +29,7 @@
#include "version_major.h"
#define LIBAVCODEC_VERSION_MINOR 34
#define LIBAVCODEC_VERSION_MINOR 35
#define LIBAVCODEC_VERSION_MICRO 100
#define LIBAVCODEC_VERSION_INT AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \