/* Copyright (C) 2002 Jean-Marc Valin    File: filters.c   Various analysis/synthesis filters   Redistribution and use in source and binary forms, with or without   modification, are permitted provided that the following conditions   are met:      - Redistributions of source code must retain the above copyright   notice, this list of conditions and the following disclaimer.      - Redistributions in binary form must reproduce the above copyright   notice, this list of conditions and the following disclaimer in the   documentation and/or other materials provided with the distribution.      - Neither the name of the Xiph.org Foundation nor the names of its   contributors may be used to endorse or promote products derived from   this software without specific prior written permission.      THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS   ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR   A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR   CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,   EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,   PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.*/#ifdef HAVE_CONFIG_H#include "config.h"#endif#include "filters.h"#include "stack_alloc.h"#include "misc.h"#include "math_approx.h"#include "ltp.h"#include <math.h>#ifdef _USE_SSE#include "filters_sse.h"#elif defined (ARM4_ASM) || defined(ARM5E_ASM)#include "filters_arm4.h"#elif defined (BFIN_ASM)#include "filters_bfin.h"#endifvoid bw_lpc(spx_word16_t gamma, const spx_coef_t *lpc_in, spx_coef_t *lpc_out, int order){   int i;   spx_word16_t tmp=gamma;   for (i=0;i<order;i++)   {      lpc_out[i] = MULT16_16_P15(tmp,lpc_in[i]);      tmp = MULT16_16_P15(tmp, gamma);   }}#ifdef FIXED_POINT/* FIXME: These functions are ugly and probably introduce too much error */void signal_mul(const spx_sig_t *x, spx_sig_t *y, spx_word32_t scale, int len){   int i;   for (i=0;i<len;i++)   {      y[i] = SHL32(MULT16_32_Q14(EXTRACT16(SHR32(x[i],7)),scale),7);   }}void signal_div(const spx_sig_t *x, spx_sig_t *y, spx_word32_t scale, int len){   int i;   if (scale > SHL32(EXTEND32(SIG_SCALING), 8))   {      spx_word16_t scale_1;      scale = PSHR32(scale, SIG_SHIFT);      scale_1 = EXTRACT16(DIV32_16(SHL32(EXTEND32(SIG_SCALING),7),scale));      for (i=0;i<len;i++)      {         y[i] = SHR32(MULT16_16(scale_1, EXTRACT16(SHR32(x[i],SIG_SHIFT))),7);      }   } else {      spx_word16_t scale_1;      scale = PSHR32(scale, SIG_SHIFT-5);      scale_1 = DIV32_16(SHL32(EXTEND32(SIG_SCALING),3),scale);      for (i=0;i<len;i++)      {         y[i] = MULT16_16(scale_1, EXTRACT16(SHR32(x[i],SIG_SHIFT-2)));      }   }}#elsevoid signal_mul(const spx_sig_t *x, spx_sig_t *y, spx_word32_t scale, int len){   int i;   for (i=0;i<len;i++)      y[i] = scale*x[i];}void signal_div(const spx_sig_t *x, spx_sig_t *y, spx_word32_t scale, int len){   int i;   float scale_1 = 1/scale;   for (i=0;i<len;i++)      y[i] = scale_1*x[i];}#endif#ifdef FIXED_POINTspx_word16_t compute_rms(const spx_sig_t *x, int len){   int i;   spx_word32_t sum=0;   spx_sig_t max_val=1;   int sig_shift;   for (i=0;i<len;i++)   {      spx_sig_t tmp = x[i];      if (tmp<0)         tmp = -tmp;      if (tmp > max_val)         max_val = tmp;   }   sig_shift=0;   while (max_val>16383)   {      sig_shift++;      max_val >>= 1;   }   for (i=0;i<len;i+=4)   {      spx_word32_t sum2=0;      spx_word16_t tmp;      tmp = EXTRACT16(SHR32(x[i],sig_shift));      sum2 = MAC16_16(sum2,tmp,tmp);      tmp = EXTRACT16(SHR32(x[i+1],sig_shift));      sum2 = MAC16_16(sum2,tmp,tmp);      tmp = EXTRACT16(SHR32(x[i+2],sig_shift));      sum2 = MAC16_16(sum2,tmp,tmp);      tmp = EXTRACT16(SHR32(x[i+3],sig_shift));      sum2 = MAC16_16(sum2,tmp,tmp);      sum = ADD32(sum,SHR32(sum2,6));   }      return EXTRACT16(SHR32(SHL32(EXTEND32(spx_sqrt(1+DIV32(sum,len))),(sig_shift+3)),SIG_SHIFT));}#ifndef OVERRIDE_NORMALIZE16int normalize16(const spx_sig_t *x, spx_word16_t *y, spx_sig_t max_scale, int len){   int i;   spx_sig_t max_val=1;   int sig_shift;      for (i=0;i<len;i++)   {      spx_sig_t tmp = x[i];      if (tmp<0)         tmp = NEG32(tmp);      if (tmp >= max_val)         max_val = tmp;   }   sig_shift=0;   while (max_val>max_scale)   {      sig_shift++;      max_val >>= 1;   }   for (i=0;i<len;i++)      y[i] = EXTRACT16(SHR32(x[i], sig_shift));      return sig_shift;}#endif#elsespx_word16_t compute_rms(const spx_sig_t *x, int len){   int i;   float sum=0;   for (i=0;i<len;i++)   {      sum += x[i]*x[i];   }   return sqrt(.1+sum/len);}#endif#ifndef OVERRIDE_FILTER_MEM2#ifdef PRECISION16void filter_mem2(const spx_sig_t *x, const spx_coef_t *num, const spx_coef_t *den, spx_sig_t *y, int N, int ord, spx_mem_t *mem){   int i,j;   spx_word16_t xi,yi,nyi;   for (i=0;i<N;i++)   {      xi= EXTRACT16(PSHR32(SATURATE(x[i],536870911),SIG_SHIFT));      yi = EXTRACT16(PSHR32(SATURATE(ADD32(x[i], SHL32(mem[0],1)),536870911),SIG_SHIFT));      nyi = NEG16(yi);      for (j=0;j<ord-1;j++)      {         mem[j] = MAC16_16(MAC16_16(mem[j+1], num[j],xi), den[j],nyi);      }      mem[ord-1] = ADD32(MULT16_16(num[ord-1],xi), MULT16_16(den[ord-1],nyi));      y[i] = SHL32(EXTEND32(yi),SIG_SHIFT);   }}#elsevoid filter_mem2(const spx_sig_t *x, const spx_coef_t *num, const spx_coef_t *den, spx_sig_t *y, int N, int ord, spx_mem_t *mem){   int i,j;   spx_sig_t xi,yi,nyi;   for (i=0;i<N;i++)   {      xi=SATURATE(x[i],805306368);      yi = SATURATE(ADD32(xi, SHL32(mem[0],2)),805306368);      nyi = NEG32(yi);      for (j=0;j<ord-1;j++)      {         mem[j] = MAC16_32_Q15(MAC16_32_Q15(mem[j+1], num[j],xi), den[j],nyi);      }      mem[ord-1] = SUB32(MULT16_32_Q15(num[ord-1],xi), MULT16_32_Q15(den[ord-1],yi));      y[i] = yi;   }}#endif#endif#ifndef OVERRIDE_IIR_MEM2#ifdef PRECISION16void iir_mem2(const spx_sig_t *x, const spx_coef_t *den, spx_sig_t *y, int N, int ord, spx_mem_t *mem){   int i,j;   spx_word16_t yi,nyi;   for (i=0;i<N;i++)   {      yi = EXTRACT16(PSHR32(SATURATE(x[i] + SHL32(mem[0],1),536870911),SIG_SHIFT));      nyi = NEG16(yi);      for (j=0;j<ord-1;j++)      {         mem[j] = MAC16_16(mem[j+1],den[j],nyi);      }      mem[ord-1] = MULT16_16(den[ord-1],nyi);      y[i] = SHL32(EXTEND32(yi),SIG_SHIFT);   }}#elsevoid iir_mem2(const spx_sig_t *x, const spx_coef_t *den, spx_sig_t *y, int N, int ord, spx_mem_t *mem){   int i,j;   spx_word32_t xi,yi,nyi;   for (i=0;i<N;i++)   {      xi=SATURATE(x[i],805306368);      yi = SATURATE(xi + SHL32(mem[0],2),805306368);      nyi = NEG32(yi);      for (j=0;j<ord-1;j++)      {         mem[j] = MAC16_32_Q15(mem[j+1],den[j],nyi);      }      mem[ord-1] = MULT16_32_Q15(den[ord-1],nyi);      y[i] = yi;   }}#endif#endif#ifndef OVERRIDE_FIR_MEM2#ifdef PRECISION16void fir_mem2(const spx_sig_t *x, const spx_coef_t *num, spx_sig_t *y, int N, int ord, spx_mem_t *mem){   int i,j;   spx_word16_t xi,yi;   for (i=0;i<N;i++)   {      xi= EXTRACT16(PSHR32(SATURATE(x[i],536870911),SIG_SHIFT));