view libfaad2/output.c @ 10730:67449e5936f3

fix 10l (computation based on uninitialized data which led to incorrect field matching) and greatly improve selection logic. the pullup core should be very accurate now, so try throwing tough samples at it and report any failures! :)

/*
** FAAD2 - Freeware Advanced Audio (AAC) Decoder including SBR decoding
** Copyright (C) 2003 M. Bakker, Ahead Software AG, http://www.nero.com
**  
** This program is free software; you can redistribute it and/or modify
** it under the terms of the GNU General Public License as published by
** the Free Software Foundation; either version 2 of the License, or
** (at your option) any later version.
** 
** This program 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 General Public License for more details.
** 
** You should have received a copy of the GNU General Public License
** along with this program; if not, write to the Free Software 
** Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
**
** Any non-GPL usage of this software or parts of this software is strictly
** forbidden.
**
** Commercial non-GPL licensing of this software is possible.
** For more info contact Ahead Software through [e-mail].
**
** $Id$
**/

#include "common.h"
#include "structs.h"

#include "output.h"
#include "decoder.h"

#ifndef FIXED_POINT

#include "dither.h"


#define ftol(A,B) {tmp = *(int32_t*) & A - 0x4B7F8000; \
                   B = (int16_t)((tmp==(int16_t)tmp) ? tmp : (tmp>>31)^0x7FFF);}

#define ROUND(x) ((x >= 0) ? (int32_t)floor((x) + 0.5) : (int32_t)ceil((x) + 0.5))

#define ROUND32(x) ROUND(x)

#define ROUND64(x) (doubletmp = (x) + Dither.Add + (int64_t)0x001FFFFD80000000L, *(int64_t*)(&doubletmp) - (int64_t)0x433FFFFD80000000L)

#define FLOAT_SCALE (1.0f/(1<<15))

dither_t Dither;
double doubletmp;

#define DM_MUL (1./(1.+sqrt(2.)))

static INLINE real_t get_sample(real_t **input, uint8_t channel, uint16_t sample,
                                uint8_t downMatrix, uint8_t *internal_channel)
{
    if (downMatrix)
    {
        if (channel == 0)
        {
            return DM_MUL * (input[internal_channel[1]][sample] +
                input[internal_channel[0]][sample]/sqrt(2.) +
                input[internal_channel[3]][sample]/sqrt(2.));
        } else {
            return DM_MUL * (input[internal_channel[2]][sample] +
                input[internal_channel[0]][sample]/sqrt(2.) +
                input[internal_channel[4]][sample]/sqrt(2.));
        }
    } else {
        return input[internal_channel[channel]][sample];
    }
}

void* output_to_PCM(faacDecHandle hDecoder,
                    real_t **input, void *sample_buffer, uint8_t channels,
                    uint16_t frame_len, uint8_t format)
{
    uint8_t ch;
    uint16_t i, j = 0;
    uint8_t internal_channel;

    int16_t   *short_sample_buffer = (int16_t*)sample_buffer;
    int32_t   *int_sample_buffer = (int32_t*)sample_buffer;
    float32_t *float_sample_buffer = (float32_t*)sample_buffer;
    double    *double_sample_buffer = (double*)sample_buffer;

    /* Copy output to a standard PCM buffer */
    for (ch = 0; ch < channels; ch++)
    {
        internal_channel = hDecoder->internal_channel[ch];

        switch (format)
        {
        case FAAD_FMT_16BIT:
            for(i = 0; i < frame_len; i++)
            {
                int32_t tmp;
                real_t ftemp;
                //real_t inp = input[internal_channel][i];
                real_t inp = get_sample(input, ch, i, hDecoder->downMatrix, hDecoder->internal_channel);

                ftemp = inp + 0xff8000;
                ftol(ftemp, short_sample_buffer[(i*channels)+ch]);
            }
            break;
        case FAAD_FMT_16BIT_DITHER:
            for(i = 0; i < frame_len; i++, j++)
            {
                real_t inp = input[internal_channel][i];
                double Sum = inp * 65535.f;
                int64_t val;
                if(j > 31)
                   j = 0;
                val = dither_output(1, 0, j, Sum, ch) / 65536;
                if (val > (1<<15)-1)
                    val = (1<<15)-1;
                else if (val < -(1<<15))
                    val = -(1<<15);
                short_sample_buffer[(i*channels)+ch] = (int16_t)val;
            }
            break;
        case FAAD_FMT_16BIT_L_SHAPE:
        case FAAD_FMT_16BIT_M_SHAPE:
        case FAAD_FMT_16BIT_H_SHAPE:
            for(i = 0; i < frame_len; i++, j++)
            {
                real_t inp = input[internal_channel][i];
                double Sum = inp * 65535.f;
                int64_t val;
                if(j > 31)
                   j = 0;
                val = dither_output(1, 1, j, Sum, ch) / 65536;
                if (val > (1<<15)-1)
                    val = (1<<15)-1;
                else if (val < -(1<<15))
                    val = -(1<<15);
                short_sample_buffer[(i*channels)+ch] = (int16_t)val;
            }
            break;
        case FAAD_FMT_24BIT:
            for(i = 0; i < frame_len; i++)
            {
                real_t inp = input[internal_channel][i];
                if (inp > (1<<15)-1)
                    inp = (1<<15)-1;
                else if (inp < -(1<<15))
                    inp = -(1<<15);
                int_sample_buffer[(i*channels)+ch] = ROUND(inp*(1<<8));
            }
            break;
        case FAAD_FMT_32BIT:
            for(i = 0; i < frame_len; i++)
            {
                real_t inp = input[internal_channel][i];
                if (inp > (1<<15)-1)
                    inp = (1<<15)-1;
                else if (inp < -(1<<15))
                    inp = -(1<<15);
                int_sample_buffer[(i*channels)+ch] = ROUND32(inp*(1<<16));
            }
            break;
        case FAAD_FMT_FLOAT:
            for(i = 0; i < frame_len; i++)
            {
                real_t inp = input[internal_channel][i];
                float_sample_buffer[(i*channels)+ch] = inp*FLOAT_SCALE;
            }
            break;
        case FAAD_FMT_DOUBLE:
            for(i = 0; i < frame_len; i++)
            {
                real_t inp = input[internal_channel][i];
                double_sample_buffer[(i*channels)+ch] = (double)inp*FLOAT_SCALE;
            }
            break;
        }
    }

    return sample_buffer;
}


/* Dither output */
static int64_t dither_output(uint8_t dithering, uint8_t shapingtype, uint16_t i, double Sum, uint8_t k)
{
    double Sum2;
    int64_t val;
    if(dithering)
    {
        if(!shapingtype)
        {
            double tmp = Random_Equi(Dither.Dither);
            Sum2 = tmp - Dither.LastRandomNumber[k];
            Dither.LastRandomNumber[k] = tmp;
            Sum2 = Sum += Sum2;
            val = ROUND64(Sum2)&Dither.Mask;
        } else {
            Sum2 = Random_Triangular(Dither.Dither) - scalar16(Dither.DitherHistory[k], Dither.FilterCoeff + i);
            Sum += Dither.DitherHistory[k][(-1-i)&15] = Sum2;
            Sum2 = Sum + scalar16(Dither.ErrorHistory[k], Dither.FilterCoeff + i );
            val = ROUND64(Sum2)&Dither.Mask;
            Dither.ErrorHistory[k][(-1-i)&15] = (float)(Sum - val);
        }
        return val;
    }
    else
        return ROUND64 (Sum);
}

#else

void* output_to_PCM(faacDecHandle hDecoder,
                    real_t **input, void *sample_buffer, uint8_t channels,
                    uint16_t frame_len, uint8_t format)
{
    uint8_t ch;
    uint16_t i;
    int16_t *short_sample_buffer = (int16_t*)sample_buffer;

    /* Copy output to a standard PCM buffer */
    for (ch = 0; ch < channels; ch++)
    {
        for(i = 0; i < frame_len; i++)
        {
            int32_t tmp = input[ch][i];
            tmp += (1 << (REAL_BITS-1));
            tmp >>= REAL_BITS;
            if (tmp > 0x7fff)       tmp = 0x7fff;
            else if (tmp <= -32768) tmp = -32768;
            short_sample_buffer[(i*channels)+ch] = (int16_t)tmp;
        }
    }

    return sample_buffer;
}

#endif