filters.c

一个开源SIP协议栈

/* Copyright (C) 2002-2006 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"
#endif



void 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);
   }
}

void highpass(const spx_word16_t *x, spx_word16_t *y, int len, int filtID, spx_mem_t *mem)
{
   int i;
#ifdef FIXED_POINT
   const spx_word16_t Pcoef[5][3] = {{16384, -31313, 14991}, {16384, -31569, 15249}, {16384, -31677, 15328}, {16384, -32313, 15947}, {16384, -22446, 6537}};
   const spx_word16_t Zcoef[5][3] = {{15672, -31344, 15672}, {15802, -31601, 15802}, {15847, -31694, 15847}, {16162, -32322, 16162}, {14418, -28836, 14418}};
#else
   const spx_word16_t Pcoef[5][3] = {{1.00000f, -1.91120f, 0.91498f}, {1.00000f, -1.92683f, 0.93071f}, {1.00000f, -1.93338f, 0.93553f}, {1.00000f, -1.97226f, 0.97332f}, {1.00000f, -1.37000f, 0.39900f}};
   const spx_word16_t Zcoef[5][3] = {{0.95654f, -1.91309f, 0.95654f}, {0.96446f, -1.92879f, 0.96446f}, {0.96723f, -1.93445f, 0.96723f}, {0.98645f, -1.97277f, 0.98645f}, {0.88000f, -1.76000f, 0.88000f}};
#endif
   const spx_word16_t *den, *num;
   if (filtID>4)
      filtID=4;
   
   den = Pcoef[filtID]; num = Zcoef[filtID];
   /*return;*/
   for (i=0;i<len;i++)
   {
      spx_word16_t yi;
      spx_word32_t vout = ADD32(MULT16_16(num[0], x[i]),mem[0]);
      yi = EXTRACT16(SATURATE(PSHR32(vout,14),32767));
      mem[0] = ADD32(MAC16_16(mem[1], num[1],x[i]), SHL32(MULT16_32_Q15(-den[1],vout),1));
      mem[1] = ADD32(MULT16_16(num[2],x[i]), SHL32(MULT16_32_Q15(-den[2],vout),1));
      y[i] = yi;
   }
}

#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_word16_t *x, spx_word16_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(PDIV32_16(SHL32(EXTEND32(SIG_SCALING),7),scale));
      for (i=0;i<len;i++)
      {
         y[i] = MULT16_16_P15(scale_1, x[i]);
      }
   } else if (scale > SHR32(EXTEND32(SIG_SCALING), 2)) {
      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] = PSHR32(MULT16_16(scale_1, SHL16(x[i],2)),8);
      }
   } else {
      spx_word16_t scale_1;
      scale = PSHR32(scale, SIG_SHIFT-7);
      if (scale < 5)
         scale = 5;
      scale_1 = DIV32_16(SHL32(EXTEND32(SIG_SCALING),3),scale);
      for (i=0;i<len;i++)
      {
         y[i] = PSHR32(MULT16_16(scale_1, SHL16(x[i],2)),6);
      }
   }
}

#else

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] = 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_POINT



spx_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(PSHR32(SHL32(EXTEND32(spx_sqrt(DIV32(sum,len))),(sig_shift+3)),SIG_SHIFT));
}

spx_word16_t compute_rms16(const spx_word16_t *x, int len)
{
   int i;
   spx_word16_t max_val=10; 

   for (i=0;i<len;i++)
   {
      spx_sig_t tmp = x[i];
      if (tmp<0)
         tmp = -tmp;
      if (tmp > max_val)
         max_val = tmp;
   }
   if (max_val>16383)
   {
      spx_word32_t sum=0;
      for (i=0;i<len;i+=4)
      {
         spx_word32_t sum2=0;
         sum2 = MAC16_16(sum2,SHR16(x[i],1),SHR16(x[i],1));
         sum2 = MAC16_16(sum2,SHR16(x[i+1],1),SHR16(x[i+1],1));
         sum2 = MAC16_16(sum2,SHR16(x[i+2],1),SHR16(x[i+2],1));
         sum2 = MAC16_16(sum2,SHR16(x[i+3],1),SHR16(x[i+3],1));
         sum = ADD32(sum,SHR32(sum2,6));
      }
      return SHL16(spx_sqrt(DIV32(sum,len)),4);
   } else {
      spx_word32_t sum=0;
      int sig_shift=0;
      if (max_val < 8192)
         sig_shift=1;
      if (max_val < 4096)
         sig_shift=2;
      if (max_val < 2048)
         sig_shift=3;
      for (i=0;i<len;i+=4)
      {
         spx_word32_t sum2=0;
         sum2 = MAC16_16(sum2,SHL16(x[i],sig_shift),SHL16(x[i],sig_shift));
         sum2 = MAC16_16(sum2,SHL16(x[i+1],sig_shift),SHL16(x[i+1],sig_shift));
         sum2 = MAC16_16(sum2,SHL16(x[i+2],sig_shift),SHL16(x[i+2],sig_shift));
         sum2 = MAC16_16(sum2,SHL16(x[i+3],sig_shift),SHL16(x[i+3],sig_shift));
         sum = ADD32(sum,SHR32(sum2,6));
      }
      return SHL16(spx_sqrt(DIV32(sum,len)),3-sig_shift);   
   }
}

#ifndef OVERRIDE_NORMALIZE16
int 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

#else

spx_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);
}
spx_word16_t compute_rms16(const spx_word16_t *x, int len)
{
   return compute_rms(x, len);
}
#endif



#ifndef OVERRIDE_FILTER_MEM2
#ifdef PRECISION16
void 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);
   }
}
#else
void 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<ord;i++)
      mem[i] = SHR32(mem[i],1);   
   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;
   }
   for (i=0;i<ord;i++)
      mem[i] = SHL32(mem[i],1);   
}
#endif
#endif

#ifdef FIXED_POINT
#ifndef OVERRIDE_FILTER_MEM16
void filter_mem16(const spx_word16_t *x, const spx_coef_t *num, const spx_coef_t *den, spx_word16_t *y, int N, int ord, spx_mem_t *mem, char *stack)
{
   int i,j;
   spx_word16_t xi,yi,nyi;
   for (i=0;i<N;i++)
   {
      xi= x[i];
      yi = EXTRACT16(SATURATE(ADD32(EXTEND32(x[i]),PSHR32(mem[0],LPC_SHIFT)),32767));
      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] = yi;
   }
}
#endif
#else
void filter_mem16(const spx_word16_t *x, const spx_coef_t *num, const spx_coef_t *den, spx_word16_t *y, int N, int ord, spx_mem_t *mem, char *stack)
{
   filter_mem2(x, num, den, y, N, ord, mem);
}
#endif


#ifndef OVERRIDE_IIR_MEM2
#ifdef PRECISION16
void 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);
   }
}
#else
void 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<ord;i++)
      mem[i] = SHR32(mem[i],1);   
   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;
   }
   for (i=0;i<ord;i++)
      mem[i] = SHL32(mem[i],1);   
}
#endif
#endif

#ifdef FIXED_POINT
#ifndef OVERRIDE_IIR_MEM16
void iir_mem16(const spx_word16_t *x, const spx_coef_t *den, spx_word16_t *y, int N, int ord, spx_mem_t *mem, char *stack)
{
   int i,j;
   spx_word16_t yi,nyi;

   for (i=0;i<N;i++)
   {
      yi = EXTRACT16(SATURATE(ADD32(EXTEND32(x[i]),PSHR32(mem[0],LPC_SHIFT)),32767));
      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] = yi;
   }
}
#endif
#else
void iir_mem16(const spx_word16_t *x, const spx_coef_t *den, spx_word16_t *y, int N, int ord, spx_mem_t *mem, char *stack)
{
   iir_mem2(x, den, y, N, ord, mem);
}
#endif


#ifndef OVERRIDE_FIR_MEM2
#ifdef PRECISION16
void 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));
      yi = EXTRACT16(PSHR32(SATURATE(x[i] + SHL32(mem[0],1),536870911),SIG_SHIFT));
      for (j=0;j<ord-1;j++)
      {
         mem[j] = MAC16_16(mem[j+1], num[j],xi);
      }
      mem[ord-1] = MULT16_16(num[ord-1],xi);
      y[i] = SHL32(EXTEND32(yi),SIG_SHIFT);
   }
}
#else