• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /* Copyright (c) 2009-2010 Xiph.Org Foundation
2    Written by Jean-Marc Valin */
3 /*
4    Redistribution and use in source and binary forms, with or without
5    modification, are permitted provided that the following conditions
6    are met:
7 
8    - Redistributions of source code must retain the above copyright
9    notice, this list of conditions and the following disclaimer.
10 
11    - Redistributions in binary form must reproduce the above copyright
12    notice, this list of conditions and the following disclaimer in the
13    documentation and/or other materials provided with the distribution.
14 
15    THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
16    ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
17    LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
18    A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
19    OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
20    EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
21    PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
22    PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
23    LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
24    NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
25    SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26 */
27 
28 #ifdef HAVE_CONFIG_H
29 #include "config.h"
30 #endif
31 
32 #include "celt_lpc.h"
33 #include "stack_alloc.h"
34 #include "mathops.h"
35 #include "pitch.h"
36 
_celt_lpc(opus_val16 * _lpc,const opus_val32 * ac,int p)37 void _celt_lpc(
38       opus_val16       *_lpc, /* out: [0...p-1] LPC coefficients      */
39 const opus_val32 *ac,  /* in:  [0...p] autocorrelation values  */
40 int          p
41 )
42 {
43    int i, j;
44    opus_val32 r;
45    opus_val32 error = ac[0];
46 #ifdef FIXED_POINT
47    opus_val32 lpc[LPC_ORDER];
48 #else
49    float *lpc = _lpc;
50 #endif
51 
52    OPUS_CLEAR(lpc, p);
53    if (ac[0] != 0)
54    {
55       for (i = 0; i < p; i++) {
56          /* Sum up this iteration's reflection coefficient */
57          opus_val32 rr = 0;
58          for (j = 0; j < i; j++)
59             rr += MULT32_32_Q31(lpc[j],ac[i - j]);
60          rr += SHR32(ac[i + 1],3);
61          r = -frac_div32(SHL32(rr,3), error);
62          /*  Update LPC coefficients and total error */
63          lpc[i] = SHR32(r,3);
64          for (j = 0; j < (i+1)>>1; j++)
65          {
66             opus_val32 tmp1, tmp2;
67             tmp1 = lpc[j];
68             tmp2 = lpc[i-1-j];
69             lpc[j]     = tmp1 + MULT32_32_Q31(r,tmp2);
70             lpc[i-1-j] = tmp2 + MULT32_32_Q31(r,tmp1);
71          }
72 
73          error = error - MULT32_32_Q31(MULT32_32_Q31(r,r),error);
74          /* Bail out once we get 30 dB gain */
75 #ifdef FIXED_POINT
76          if (error<SHR32(ac[0],10))
77             break;
78 #else
79          if (error<.001f*ac[0])
80             break;
81 #endif
82       }
83    }
84 #ifdef FIXED_POINT
85    for (i=0;i<p;i++)
86       _lpc[i] = ROUND16(lpc[i],16);
87 #endif
88 }
89 
90 
celt_fir_c(const opus_val16 * _x,const opus_val16 * num,opus_val16 * _y,int N,int ord,opus_val16 * mem,int arch)91 void celt_fir_c(
92          const opus_val16 *_x,
93          const opus_val16 *num,
94          opus_val16 *_y,
95          int N,
96          int ord,
97          opus_val16 *mem,
98          int arch)
99 {
100    int i,j;
101    VARDECL(opus_val16, rnum);
102    VARDECL(opus_val16, x);
103    SAVE_STACK;
104 
105    ALLOC(rnum, ord, opus_val16);
106    ALLOC(x, N+ord, opus_val16);
107    for(i=0;i<ord;i++)
108       rnum[i] = num[ord-i-1];
109    for(i=0;i<ord;i++)
110       x[i] = mem[ord-i-1];
111    for (i=0;i<N;i++)
112       x[i+ord]=_x[i];
113    for(i=0;i<ord;i++)
114       mem[i] = _x[N-i-1];
115 #ifdef SMALL_FOOTPRINT
116    (void)arch;
117    for (i=0;i<N;i++)
118    {
119       opus_val32 sum = SHL32(EXTEND32(_x[i]), SIG_SHIFT);
120       for (j=0;j<ord;j++)
121       {
122          sum = MAC16_16(sum,rnum[j],x[i+j]);
123       }
124       _y[i] = SATURATE16(PSHR32(sum, SIG_SHIFT));
125    }
126 #else
127    for (i=0;i<N-3;i+=4)
128    {
129       opus_val32 sum[4]={0,0,0,0};
130       xcorr_kernel(rnum, x+i, sum, ord, arch);
131       _y[i  ] = SATURATE16(ADD32(EXTEND32(_x[i  ]), PSHR32(sum[0], SIG_SHIFT)));
132       _y[i+1] = SATURATE16(ADD32(EXTEND32(_x[i+1]), PSHR32(sum[1], SIG_SHIFT)));
133       _y[i+2] = SATURATE16(ADD32(EXTEND32(_x[i+2]), PSHR32(sum[2], SIG_SHIFT)));
134       _y[i+3] = SATURATE16(ADD32(EXTEND32(_x[i+3]), PSHR32(sum[3], SIG_SHIFT)));
135    }
136    for (;i<N;i++)
137    {
138       opus_val32 sum = 0;
139       for (j=0;j<ord;j++)
140          sum = MAC16_16(sum,rnum[j],x[i+j]);
141       _y[i] = SATURATE16(ADD32(EXTEND32(_x[i]), PSHR32(sum, SIG_SHIFT)));
142    }
143 #endif
144    RESTORE_STACK;
145 }
146 
celt_iir(const opus_val32 * _x,const opus_val16 * den,opus_val32 * _y,int N,int ord,opus_val16 * mem,int arch)147 void celt_iir(const opus_val32 *_x,
148          const opus_val16 *den,
149          opus_val32 *_y,
150          int N,
151          int ord,
152          opus_val16 *mem,
153          int arch)
154 {
155 #ifdef SMALL_FOOTPRINT
156    int i,j;
157    (void)arch;
158    for (i=0;i<N;i++)
159    {
160       opus_val32 sum = _x[i];
161       for (j=0;j<ord;j++)
162       {
163          sum -= MULT16_16(den[j],mem[j]);
164       }
165       for (j=ord-1;j>=1;j--)
166       {
167          mem[j]=mem[j-1];
168       }
169       mem[0] = ROUND16(sum,SIG_SHIFT);
170       _y[i] = sum;
171    }
172 #else
173    int i,j;
174    VARDECL(opus_val16, rden);
175    VARDECL(opus_val16, y);
176    SAVE_STACK;
177 
178    celt_assert((ord&3)==0);
179    ALLOC(rden, ord, opus_val16);
180    ALLOC(y, N+ord, opus_val16);
181    for(i=0;i<ord;i++)
182       rden[i] = den[ord-i-1];
183    for(i=0;i<ord;i++)
184       y[i] = -mem[ord-i-1];
185    for(;i<N+ord;i++)
186       y[i]=0;
187    for (i=0;i<N-3;i+=4)
188    {
189       /* Unroll by 4 as if it were an FIR filter */
190       opus_val32 sum[4];
191       sum[0]=_x[i];
192       sum[1]=_x[i+1];
193       sum[2]=_x[i+2];
194       sum[3]=_x[i+3];
195       xcorr_kernel(rden, y+i, sum, ord, arch);
196 
197       /* Patch up the result to compensate for the fact that this is an IIR */
198       y[i+ord  ] = -ROUND16(sum[0],SIG_SHIFT);
199       _y[i  ] = sum[0];
200       sum[1] = MAC16_16(sum[1], y[i+ord  ], den[0]);
201       y[i+ord+1] = -ROUND16(sum[1],SIG_SHIFT);
202       _y[i+1] = sum[1];
203       sum[2] = MAC16_16(sum[2], y[i+ord+1], den[0]);
204       sum[2] = MAC16_16(sum[2], y[i+ord  ], den[1]);
205       y[i+ord+2] = -ROUND16(sum[2],SIG_SHIFT);
206       _y[i+2] = sum[2];
207 
208       sum[3] = MAC16_16(sum[3], y[i+ord+2], den[0]);
209       sum[3] = MAC16_16(sum[3], y[i+ord+1], den[1]);
210       sum[3] = MAC16_16(sum[3], y[i+ord  ], den[2]);
211       y[i+ord+3] = -ROUND16(sum[3],SIG_SHIFT);
212       _y[i+3] = sum[3];
213    }
214    for (;i<N;i++)
215    {
216       opus_val32 sum = _x[i];
217       for (j=0;j<ord;j++)
218          sum -= MULT16_16(rden[j],y[i+j]);
219       y[i+ord] = ROUND16(sum,SIG_SHIFT);
220       _y[i] = sum;
221    }
222    for(i=0;i<ord;i++)
223       mem[i] = _y[N-i-1];
224    RESTORE_STACK;
225 #endif
226 }
227 
_celt_autocorr(const opus_val16 * x,opus_val32 * ac,const opus_val16 * window,int overlap,int lag,int n,int arch)228 int _celt_autocorr(
229                    const opus_val16 *x,   /*  in: [0...n-1] samples x   */
230                    opus_val32       *ac,  /* out: [0...lag-1] ac values */
231                    const opus_val16       *window,
232                    int          overlap,
233                    int          lag,
234                    int          n,
235                    int          arch
236                   )
237 {
238    opus_val32 d;
239    int i, k;
240    int fastN=n-lag;
241    int shift;
242    const opus_val16 *xptr;
243    VARDECL(opus_val16, xx);
244    SAVE_STACK;
245    ALLOC(xx, n, opus_val16);
246    celt_assert(n>0);
247    celt_assert(overlap>=0);
248    if (overlap == 0)
249    {
250       xptr = x;
251    } else {
252       for (i=0;i<n;i++)
253          xx[i] = x[i];
254       for (i=0;i<overlap;i++)
255       {
256          xx[i] = MULT16_16_Q15(x[i],window[i]);
257          xx[n-i-1] = MULT16_16_Q15(x[n-i-1],window[i]);
258       }
259       xptr = xx;
260    }
261    shift=0;
262 #ifdef FIXED_POINT
263    {
264       opus_val32 ac0;
265       ac0 = 1+(n<<7);
266       if (n&1) ac0 += SHR32(MULT16_16(xptr[0],xptr[0]),9);
267       for(i=(n&1);i<n;i+=2)
268       {
269          ac0 += SHR32(MULT16_16(xptr[i],xptr[i]),9);
270          ac0 += SHR32(MULT16_16(xptr[i+1],xptr[i+1]),9);
271       }
272 
273       shift = celt_ilog2(ac0)-30+10;
274       shift = (shift)/2;
275       if (shift>0)
276       {
277          for(i=0;i<n;i++)
278             xx[i] = PSHR32(xptr[i], shift);
279          xptr = xx;
280       } else
281          shift = 0;
282    }
283 #endif
284    celt_pitch_xcorr(xptr, xptr, ac, fastN, lag+1, arch);
285    for (k=0;k<=lag;k++)
286    {
287       for (i = k+fastN, d = 0; i < n; i++)
288          d = MAC16_16(d, xptr[i], xptr[i-k]);
289       ac[k] += d;
290    }
291 #ifdef FIXED_POINT
292    shift = 2*shift;
293    if (shift<=0)
294       ac[0] += SHL32((opus_int32)1, -shift);
295    if (ac[0] < 268435456)
296    {
297       int shift2 = 29 - EC_ILOG(ac[0]);
298       for (i=0;i<=lag;i++)
299          ac[i] = SHL32(ac[i], shift2);
300       shift -= shift2;
301    } else if (ac[0] >= 536870912)
302    {
303       int shift2=1;
304       if (ac[0] >= 1073741824)
305          shift2++;
306       for (i=0;i<=lag;i++)
307          ac[i] = SHR32(ac[i], shift2);
308       shift += shift2;
309    }
310 #endif
311 
312    RESTORE_STACK;
313    return shift;
314 }
315