mirror of
https://github.com/xenia-project/FFmpeg.git
synced 2024-11-24 12:09:55 +00:00
b5f09d31c2
.data section. Originally committed as revision 19787 to svn://svn.ffmpeg.org/ffmpeg/trunk
579 lines
19 KiB
C
579 lines
19 KiB
C
/*
|
|
* PCM codecs
|
|
* Copyright (c) 2001 Fabrice Bellard
|
|
*
|
|
* 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
|
|
*/
|
|
|
|
/**
|
|
* @file libavcodec/pcm.c
|
|
* PCM codecs
|
|
*/
|
|
|
|
#include "avcodec.h"
|
|
#include "get_bits.h" // for ff_reverse
|
|
#include "bytestream.h"
|
|
|
|
#define MAX_CHANNELS 64
|
|
|
|
/* from g711.c by SUN microsystems (unrestricted use) */
|
|
|
|
#define SIGN_BIT (0x80) /* Sign bit for a A-law byte. */
|
|
#define QUANT_MASK (0xf) /* Quantization field mask. */
|
|
#define NSEGS (8) /* Number of A-law segments. */
|
|
#define SEG_SHIFT (4) /* Left shift for segment number. */
|
|
#define SEG_MASK (0x70) /* Segment field mask. */
|
|
|
|
#define BIAS (0x84) /* Bias for linear code. */
|
|
|
|
/*
|
|
* alaw2linear() - Convert an A-law value to 16-bit linear PCM
|
|
*
|
|
*/
|
|
static av_cold int alaw2linear(unsigned char a_val)
|
|
{
|
|
int t;
|
|
int seg;
|
|
|
|
a_val ^= 0x55;
|
|
|
|
t = a_val & QUANT_MASK;
|
|
seg = ((unsigned)a_val & SEG_MASK) >> SEG_SHIFT;
|
|
if(seg) t= (t + t + 1 + 32) << (seg + 2);
|
|
else t= (t + t + 1 ) << 3;
|
|
|
|
return (a_val & SIGN_BIT) ? t : -t;
|
|
}
|
|
|
|
static av_cold int ulaw2linear(unsigned char u_val)
|
|
{
|
|
int t;
|
|
|
|
/* Complement to obtain normal u-law value. */
|
|
u_val = ~u_val;
|
|
|
|
/*
|
|
* Extract and bias the quantization bits. Then
|
|
* shift up by the segment number and subtract out the bias.
|
|
*/
|
|
t = ((u_val & QUANT_MASK) << 3) + BIAS;
|
|
t <<= ((unsigned)u_val & SEG_MASK) >> SEG_SHIFT;
|
|
|
|
return (u_val & SIGN_BIT) ? (BIAS - t) : (t - BIAS);
|
|
}
|
|
|
|
/* 16384 entries per table */
|
|
static uint8_t linear_to_alaw[16384];
|
|
static uint8_t linear_to_ulaw[16384];
|
|
|
|
static av_cold void build_xlaw_table(uint8_t *linear_to_xlaw,
|
|
int (*xlaw2linear)(unsigned char),
|
|
int mask)
|
|
{
|
|
int i, j, v, v1, v2;
|
|
|
|
j = 0;
|
|
for(i=0;i<128;i++) {
|
|
if (i != 127) {
|
|
v1 = xlaw2linear(i ^ mask);
|
|
v2 = xlaw2linear((i + 1) ^ mask);
|
|
v = (v1 + v2 + 4) >> 3;
|
|
} else {
|
|
v = 8192;
|
|
}
|
|
for(;j<v;j++) {
|
|
linear_to_xlaw[8192 + j] = (i ^ mask);
|
|
if (j > 0)
|
|
linear_to_xlaw[8192 - j] = (i ^ (mask ^ 0x80));
|
|
}
|
|
}
|
|
linear_to_xlaw[0] = linear_to_xlaw[1];
|
|
}
|
|
|
|
static av_cold int pcm_encode_init(AVCodecContext *avctx)
|
|
{
|
|
avctx->frame_size = 1;
|
|
switch(avctx->codec->id) {
|
|
case CODEC_ID_PCM_ALAW:
|
|
build_xlaw_table(linear_to_alaw, alaw2linear, 0xd5);
|
|
break;
|
|
case CODEC_ID_PCM_MULAW:
|
|
build_xlaw_table(linear_to_ulaw, ulaw2linear, 0xff);
|
|
break;
|
|
default:
|
|
break;
|
|
}
|
|
|
|
avctx->bits_per_coded_sample = av_get_bits_per_sample(avctx->codec->id);
|
|
avctx->block_align = avctx->channels * avctx->bits_per_coded_sample/8;
|
|
avctx->coded_frame= avcodec_alloc_frame();
|
|
avctx->coded_frame->key_frame= 1;
|
|
|
|
return 0;
|
|
}
|
|
|
|
static av_cold int pcm_encode_close(AVCodecContext *avctx)
|
|
{
|
|
av_freep(&avctx->coded_frame);
|
|
|
|
return 0;
|
|
}
|
|
|
|
/**
|
|
* Write PCM samples macro
|
|
* @param type Datatype of native machine format
|
|
* @param endian bytestream_put_xxx() suffix
|
|
* @param src Source pointer (variable name)
|
|
* @param dst Destination pointer (variable name)
|
|
* @param n Total number of samples (variable name)
|
|
* @param shift Bitshift (bits)
|
|
* @param offset Sample value offset
|
|
*/
|
|
#define ENCODE(type, endian, src, dst, n, shift, offset) \
|
|
samples_##type = (type*)src; \
|
|
for(;n>0;n--) { \
|
|
register type v = (*samples_##type++ >> shift) + offset; \
|
|
bytestream_put_##endian(&dst, v); \
|
|
}
|
|
|
|
static int pcm_encode_frame(AVCodecContext *avctx,
|
|
unsigned char *frame, int buf_size, void *data)
|
|
{
|
|
int n, sample_size, v;
|
|
short *samples;
|
|
unsigned char *dst;
|
|
uint8_t *srcu8;
|
|
int16_t *samples_int16_t;
|
|
int32_t *samples_int32_t;
|
|
int64_t *samples_int64_t;
|
|
uint16_t *samples_uint16_t;
|
|
uint32_t *samples_uint32_t;
|
|
|
|
sample_size = av_get_bits_per_sample(avctx->codec->id)/8;
|
|
n = buf_size / sample_size;
|
|
samples = data;
|
|
dst = frame;
|
|
|
|
if (avctx->sample_fmt!=avctx->codec->sample_fmts[0]) {
|
|
av_log(avctx, AV_LOG_ERROR, "invalid sample_fmt\n");
|
|
return -1;
|
|
}
|
|
|
|
switch(avctx->codec->id) {
|
|
case CODEC_ID_PCM_U32LE:
|
|
ENCODE(uint32_t, le32, samples, dst, n, 0, 0x80000000)
|
|
break;
|
|
case CODEC_ID_PCM_U32BE:
|
|
ENCODE(uint32_t, be32, samples, dst, n, 0, 0x80000000)
|
|
break;
|
|
case CODEC_ID_PCM_S24LE:
|
|
ENCODE(int32_t, le24, samples, dst, n, 8, 0)
|
|
break;
|
|
case CODEC_ID_PCM_S24BE:
|
|
ENCODE(int32_t, be24, samples, dst, n, 8, 0)
|
|
break;
|
|
case CODEC_ID_PCM_U24LE:
|
|
ENCODE(uint32_t, le24, samples, dst, n, 8, 0x800000)
|
|
break;
|
|
case CODEC_ID_PCM_U24BE:
|
|
ENCODE(uint32_t, be24, samples, dst, n, 8, 0x800000)
|
|
break;
|
|
case CODEC_ID_PCM_S24DAUD:
|
|
for(;n>0;n--) {
|
|
uint32_t tmp = ff_reverse[(*samples >> 8) & 0xff] +
|
|
(ff_reverse[*samples & 0xff] << 8);
|
|
tmp <<= 4; // sync flags would go here
|
|
bytestream_put_be24(&dst, tmp);
|
|
samples++;
|
|
}
|
|
break;
|
|
case CODEC_ID_PCM_U16LE:
|
|
ENCODE(uint16_t, le16, samples, dst, n, 0, 0x8000)
|
|
break;
|
|
case CODEC_ID_PCM_U16BE:
|
|
ENCODE(uint16_t, be16, samples, dst, n, 0, 0x8000)
|
|
break;
|
|
case CODEC_ID_PCM_S8:
|
|
srcu8= data;
|
|
for(;n>0;n--) {
|
|
v = *srcu8++;
|
|
*dst++ = v - 128;
|
|
}
|
|
break;
|
|
#if HAVE_BIGENDIAN
|
|
case CODEC_ID_PCM_F64LE:
|
|
ENCODE(int64_t, le64, samples, dst, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_S32LE:
|
|
case CODEC_ID_PCM_F32LE:
|
|
ENCODE(int32_t, le32, samples, dst, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_S16LE:
|
|
ENCODE(int16_t, le16, samples, dst, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_F64BE:
|
|
case CODEC_ID_PCM_F32BE:
|
|
case CODEC_ID_PCM_S32BE:
|
|
case CODEC_ID_PCM_S16BE:
|
|
#else
|
|
case CODEC_ID_PCM_F64BE:
|
|
ENCODE(int64_t, be64, samples, dst, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_F32BE:
|
|
case CODEC_ID_PCM_S32BE:
|
|
ENCODE(int32_t, be32, samples, dst, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_S16BE:
|
|
ENCODE(int16_t, be16, samples, dst, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_F64LE:
|
|
case CODEC_ID_PCM_F32LE:
|
|
case CODEC_ID_PCM_S32LE:
|
|
case CODEC_ID_PCM_S16LE:
|
|
#endif /* HAVE_BIGENDIAN */
|
|
case CODEC_ID_PCM_U8:
|
|
memcpy(dst, samples, n*sample_size);
|
|
dst += n*sample_size;
|
|
break;
|
|
case CODEC_ID_PCM_ZORK:
|
|
for(;n>0;n--) {
|
|
v= *samples++ >> 8;
|
|
if(v<0) v = -v;
|
|
else v+= 128;
|
|
*dst++ = v;
|
|
}
|
|
break;
|
|
case CODEC_ID_PCM_ALAW:
|
|
for(;n>0;n--) {
|
|
v = *samples++;
|
|
*dst++ = linear_to_alaw[(v + 32768) >> 2];
|
|
}
|
|
break;
|
|
case CODEC_ID_PCM_MULAW:
|
|
for(;n>0;n--) {
|
|
v = *samples++;
|
|
*dst++ = linear_to_ulaw[(v + 32768) >> 2];
|
|
}
|
|
break;
|
|
default:
|
|
return -1;
|
|
}
|
|
//avctx->frame_size = (dst - frame) / (sample_size * avctx->channels);
|
|
|
|
return dst - frame;
|
|
}
|
|
|
|
typedef struct PCMDecode {
|
|
short table[256];
|
|
} PCMDecode;
|
|
|
|
static av_cold int pcm_decode_init(AVCodecContext * avctx)
|
|
{
|
|
PCMDecode *s = avctx->priv_data;
|
|
int i;
|
|
|
|
switch(avctx->codec->id) {
|
|
case CODEC_ID_PCM_ALAW:
|
|
for(i=0;i<256;i++)
|
|
s->table[i] = alaw2linear(i);
|
|
break;
|
|
case CODEC_ID_PCM_MULAW:
|
|
for(i=0;i<256;i++)
|
|
s->table[i] = ulaw2linear(i);
|
|
break;
|
|
default:
|
|
break;
|
|
}
|
|
|
|
avctx->sample_fmt = avctx->codec->sample_fmts[0];
|
|
return 0;
|
|
}
|
|
|
|
/**
|
|
* Read PCM samples macro
|
|
* @param type Datatype of native machine format
|
|
* @param endian bytestream_get_xxx() endian suffix
|
|
* @param src Source pointer (variable name)
|
|
* @param dst Destination pointer (variable name)
|
|
* @param n Total number of samples (variable name)
|
|
* @param shift Bitshift (bits)
|
|
* @param offset Sample value offset
|
|
*/
|
|
#define DECODE(type, endian, src, dst, n, shift, offset) \
|
|
dst_##type = (type*)dst; \
|
|
for(;n>0;n--) { \
|
|
register type v = bytestream_get_##endian(&src); \
|
|
*dst_##type++ = (v - offset) << shift; \
|
|
} \
|
|
dst = (short*)dst_##type;
|
|
|
|
static int pcm_decode_frame(AVCodecContext *avctx,
|
|
void *data, int *data_size,
|
|
AVPacket *avpkt)
|
|
{
|
|
const uint8_t *buf = avpkt->data;
|
|
int buf_size = avpkt->size;
|
|
PCMDecode *s = avctx->priv_data;
|
|
int sample_size, c, n;
|
|
short *samples;
|
|
const uint8_t *src, *src8, *src2[MAX_CHANNELS];
|
|
uint8_t *dstu8;
|
|
int16_t *dst_int16_t;
|
|
int32_t *dst_int32_t;
|
|
int64_t *dst_int64_t;
|
|
uint16_t *dst_uint16_t;
|
|
uint32_t *dst_uint32_t;
|
|
|
|
samples = data;
|
|
src = buf;
|
|
|
|
if (avctx->sample_fmt!=avctx->codec->sample_fmts[0]) {
|
|
av_log(avctx, AV_LOG_ERROR, "invalid sample_fmt\n");
|
|
return -1;
|
|
}
|
|
|
|
if(avctx->channels <= 0 || avctx->channels > MAX_CHANNELS){
|
|
av_log(avctx, AV_LOG_ERROR, "PCM channels out of bounds\n");
|
|
return -1;
|
|
}
|
|
|
|
sample_size = av_get_bits_per_sample(avctx->codec_id)/8;
|
|
|
|
/* av_get_bits_per_sample returns 0 for CODEC_ID_PCM_DVD */
|
|
if (CODEC_ID_PCM_DVD == avctx->codec_id)
|
|
/* 2 samples are interleaved per block in PCM_DVD */
|
|
sample_size = avctx->bits_per_coded_sample * 2 / 8;
|
|
|
|
n = avctx->channels * sample_size;
|
|
|
|
if(n && buf_size % n){
|
|
if (buf_size < n) {
|
|
av_log(avctx, AV_LOG_ERROR, "invalid PCM packet\n");
|
|
return -1;
|
|
}else
|
|
buf_size -= buf_size % n;
|
|
}
|
|
|
|
buf_size= FFMIN(buf_size, *data_size/2);
|
|
*data_size=0;
|
|
|
|
n = buf_size/sample_size;
|
|
|
|
switch(avctx->codec->id) {
|
|
case CODEC_ID_PCM_U32LE:
|
|
DECODE(uint32_t, le32, src, samples, n, 0, 0x80000000)
|
|
break;
|
|
case CODEC_ID_PCM_U32BE:
|
|
DECODE(uint32_t, be32, src, samples, n, 0, 0x80000000)
|
|
break;
|
|
case CODEC_ID_PCM_S24LE:
|
|
DECODE(int32_t, le24, src, samples, n, 8, 0)
|
|
break;
|
|
case CODEC_ID_PCM_S24BE:
|
|
DECODE(int32_t, be24, src, samples, n, 8, 0)
|
|
break;
|
|
case CODEC_ID_PCM_U24LE:
|
|
DECODE(uint32_t, le24, src, samples, n, 8, 0x800000)
|
|
break;
|
|
case CODEC_ID_PCM_U24BE:
|
|
DECODE(uint32_t, be24, src, samples, n, 8, 0x800000)
|
|
break;
|
|
case CODEC_ID_PCM_S24DAUD:
|
|
for(;n>0;n--) {
|
|
uint32_t v = bytestream_get_be24(&src);
|
|
v >>= 4; // sync flags are here
|
|
*samples++ = ff_reverse[(v >> 8) & 0xff] +
|
|
(ff_reverse[v & 0xff] << 8);
|
|
}
|
|
break;
|
|
case CODEC_ID_PCM_S16LE_PLANAR:
|
|
n /= avctx->channels;
|
|
for(c=0;c<avctx->channels;c++)
|
|
src2[c] = &src[c*n*2];
|
|
for(;n>0;n--)
|
|
for(c=0;c<avctx->channels;c++)
|
|
*samples++ = bytestream_get_le16(&src2[c]);
|
|
src = src2[avctx->channels-1];
|
|
break;
|
|
case CODEC_ID_PCM_U16LE:
|
|
DECODE(uint16_t, le16, src, samples, n, 0, 0x8000)
|
|
break;
|
|
case CODEC_ID_PCM_U16BE:
|
|
DECODE(uint16_t, be16, src, samples, n, 0, 0x8000)
|
|
break;
|
|
case CODEC_ID_PCM_S8:
|
|
dstu8= (uint8_t*)samples;
|
|
for(;n>0;n--) {
|
|
*dstu8++ = *src++ + 128;
|
|
}
|
|
samples= (short*)dstu8;
|
|
break;
|
|
#if HAVE_BIGENDIAN
|
|
case CODEC_ID_PCM_F64LE:
|
|
DECODE(int64_t, le64, src, samples, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_S32LE:
|
|
case CODEC_ID_PCM_F32LE:
|
|
DECODE(int32_t, le32, src, samples, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_S16LE:
|
|
DECODE(int16_t, le16, src, samples, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_F64BE:
|
|
case CODEC_ID_PCM_F32BE:
|
|
case CODEC_ID_PCM_S32BE:
|
|
case CODEC_ID_PCM_S16BE:
|
|
#else
|
|
case CODEC_ID_PCM_F64BE:
|
|
DECODE(int64_t, be64, src, samples, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_F32BE:
|
|
case CODEC_ID_PCM_S32BE:
|
|
DECODE(int32_t, be32, src, samples, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_S16BE:
|
|
DECODE(int16_t, be16, src, samples, n, 0, 0)
|
|
break;
|
|
case CODEC_ID_PCM_F64LE:
|
|
case CODEC_ID_PCM_F32LE:
|
|
case CODEC_ID_PCM_S32LE:
|
|
case CODEC_ID_PCM_S16LE:
|
|
#endif /* HAVE_BIGENDIAN */
|
|
case CODEC_ID_PCM_U8:
|
|
memcpy(samples, src, n*sample_size);
|
|
src += n*sample_size;
|
|
samples = (short*)((uint8_t*)data + n*sample_size);
|
|
break;
|
|
case CODEC_ID_PCM_ZORK:
|
|
for(;n>0;n--) {
|
|
int x= *src++;
|
|
if(x&128) x-= 128;
|
|
else x = -x;
|
|
*samples++ = x << 8;
|
|
}
|
|
break;
|
|
case CODEC_ID_PCM_ALAW:
|
|
case CODEC_ID_PCM_MULAW:
|
|
for(;n>0;n--) {
|
|
*samples++ = s->table[*src++];
|
|
}
|
|
break;
|
|
case CODEC_ID_PCM_DVD:
|
|
dst_int32_t = data;
|
|
n /= avctx->channels;
|
|
switch (avctx->bits_per_coded_sample) {
|
|
case 20:
|
|
while (n--) {
|
|
c = avctx->channels;
|
|
src8 = src + 4*c;
|
|
while (c--) {
|
|
*dst_int32_t++ = (bytestream_get_be16(&src) << 16) + ((*src8 &0xf0) << 8);
|
|
*dst_int32_t++ = (bytestream_get_be16(&src) << 16) + ((*src8++ &0x0f) << 12);
|
|
}
|
|
src = src8;
|
|
}
|
|
break;
|
|
case 24:
|
|
while (n--) {
|
|
c = avctx->channels;
|
|
src8 = src + 4*c;
|
|
while (c--) {
|
|
*dst_int32_t++ = (bytestream_get_be16(&src) << 16) + ((*src8++) << 8);
|
|
*dst_int32_t++ = (bytestream_get_be16(&src) << 16) + ((*src8++) << 8);
|
|
}
|
|
src = src8;
|
|
}
|
|
break;
|
|
default:
|
|
av_log(avctx, AV_LOG_ERROR, "PCM DVD unsupported sample depth\n");
|
|
return -1;
|
|
break;
|
|
}
|
|
samples = (short *) dst_int32_t;
|
|
break;
|
|
default:
|
|
return -1;
|
|
}
|
|
*data_size = (uint8_t *)samples - (uint8_t *)data;
|
|
return src - buf;
|
|
}
|
|
|
|
#if CONFIG_ENCODERS
|
|
#define PCM_ENCODER(id,sample_fmt_,name,long_name_) \
|
|
AVCodec name ## _encoder = { \
|
|
#name, \
|
|
CODEC_TYPE_AUDIO, \
|
|
id, \
|
|
0, \
|
|
pcm_encode_init, \
|
|
pcm_encode_frame, \
|
|
pcm_encode_close, \
|
|
NULL, \
|
|
.sample_fmts = (const enum SampleFormat[]){sample_fmt_,SAMPLE_FMT_NONE}, \
|
|
.long_name = NULL_IF_CONFIG_SMALL(long_name_), \
|
|
};
|
|
#else
|
|
#define PCM_ENCODER(id,sample_fmt_,name,long_name_)
|
|
#endif
|
|
|
|
#if CONFIG_DECODERS
|
|
#define PCM_DECODER(id,sample_fmt_,name,long_name_) \
|
|
AVCodec name ## _decoder = { \
|
|
#name, \
|
|
CODEC_TYPE_AUDIO, \
|
|
id, \
|
|
sizeof(PCMDecode), \
|
|
pcm_decode_init, \
|
|
NULL, \
|
|
NULL, \
|
|
pcm_decode_frame, \
|
|
.sample_fmts = (const enum SampleFormat[]){sample_fmt_,SAMPLE_FMT_NONE}, \
|
|
.long_name = NULL_IF_CONFIG_SMALL(long_name_), \
|
|
};
|
|
#else
|
|
#define PCM_DECODER(id,sample_fmt_,name,long_name_)
|
|
#endif
|
|
|
|
#define PCM_CODEC(id, sample_fmt_, name, long_name_) \
|
|
PCM_ENCODER(id,sample_fmt_,name,long_name_) PCM_DECODER(id,sample_fmt_,name,long_name_)
|
|
|
|
/* Note: Do not forget to add new entries to the Makefile as well. */
|
|
PCM_CODEC (CODEC_ID_PCM_ALAW, SAMPLE_FMT_S16, pcm_alaw, "PCM A-law");
|
|
PCM_CODEC (CODEC_ID_PCM_DVD, SAMPLE_FMT_S32, pcm_dvd, "PCM signed 20|24-bit big-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_F32BE, SAMPLE_FMT_FLT, pcm_f32be, "PCM 32-bit floating point big-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_F32LE, SAMPLE_FMT_FLT, pcm_f32le, "PCM 32-bit floating point little-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_F64BE, SAMPLE_FMT_DBL, pcm_f64be, "PCM 64-bit floating point big-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_F64LE, SAMPLE_FMT_DBL, pcm_f64le, "PCM 64-bit floating point little-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_MULAW, SAMPLE_FMT_S16, pcm_mulaw, "PCM mu-law");
|
|
PCM_CODEC (CODEC_ID_PCM_S8, SAMPLE_FMT_U8, pcm_s8, "PCM signed 8-bit");
|
|
PCM_CODEC (CODEC_ID_PCM_S16BE, SAMPLE_FMT_S16, pcm_s16be, "PCM signed 16-bit big-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_S16LE, SAMPLE_FMT_S16, pcm_s16le, "PCM signed 16-bit little-endian");
|
|
PCM_DECODER(CODEC_ID_PCM_S16LE_PLANAR, SAMPLE_FMT_S16, pcm_s16le_planar, "PCM 16-bit little-endian planar");
|
|
PCM_CODEC (CODEC_ID_PCM_S24BE, SAMPLE_FMT_S32, pcm_s24be, "PCM signed 24-bit big-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_S24DAUD, SAMPLE_FMT_S16, pcm_s24daud, "PCM D-Cinema audio signed 24-bit");
|
|
PCM_CODEC (CODEC_ID_PCM_S24LE, SAMPLE_FMT_S32, pcm_s24le, "PCM signed 24-bit little-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_S32BE, SAMPLE_FMT_S32, pcm_s32be, "PCM signed 32-bit big-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_S32LE, SAMPLE_FMT_S32, pcm_s32le, "PCM signed 32-bit little-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_U8, SAMPLE_FMT_U8, pcm_u8, "PCM unsigned 8-bit");
|
|
PCM_CODEC (CODEC_ID_PCM_U16BE, SAMPLE_FMT_S16, pcm_u16be, "PCM unsigned 16-bit big-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_U16LE, SAMPLE_FMT_S16, pcm_u16le, "PCM unsigned 16-bit little-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_U24BE, SAMPLE_FMT_S32, pcm_u24be, "PCM unsigned 24-bit big-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_U24LE, SAMPLE_FMT_S32, pcm_u24le, "PCM unsigned 24-bit little-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_U32BE, SAMPLE_FMT_S32, pcm_u32be, "PCM unsigned 32-bit big-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_U32LE, SAMPLE_FMT_S32, pcm_u32le, "PCM unsigned 32-bit little-endian");
|
|
PCM_CODEC (CODEC_ID_PCM_ZORK, SAMPLE_FMT_S16, pcm_zork, "PCM Zork");
|