1 /********************************************************************
2  *                                                                  *
3  * THIS FILE IS PART OF THE OggVorbis SOFTWARE CODEC SOURCE CODE.   *
4  * USE, DISTRIBUTION AND REPRODUCTION OF THIS LIBRARY SOURCE IS     *
5  * GOVERNED BY A BSD-STYLE SOURCE LICENSE INCLUDED WITH THIS SOURCE *
6  * IN 'COPYING'. PLEASE READ THESE TERMS BEFORE DISTRIBUTING.       *
7  *                                                                  *
8  * THE OggVorbis SOURCE CODE IS (C) COPYRIGHT 1994-2002             *
9  * by the XIPHOPHORUS Company http://www.xiph.org/                  *
10  *                                                                  *
11  ********************************************************************
12 
13   function: LSP (also called LSF) conversion routines
14   last mod: $Id: lsp.c 7187 2004-07-20 07:24:27Z xiphmont $
15 
16   The LSP generation code is taken (with minimal modification and a
17   few bugfixes) from "On the Computation of the LSP Frequencies" by
18   Joseph Rothweiler (see http://www.rothweiler.us for contact info).
19   The paper is available at:
20 
21   http://www.myown1.com/joe/lsf
22 
23  ********************************************************************/
24 
25 /* Note that the lpc-lsp conversion finds the roots of polynomial with
26    an iterative root polisher (CACM algorithm 283).  It *is* possible
27    to confuse this algorithm into not converging; that should only
28    happen with absurdly closely spaced roots (very sharp peaks in the
29    LPC f response) which in turn should be impossible in our use of
30    the code.  If this *does* happen anyway, it's a bug in the floor
31    finder; find the cause of the confusion (probably a single bin
32    spike or accidental near-float-limit resolution problems) and
33    correct it. */
34 
35 #include <math.h>
36 #include <string.h>
37 #include <stdlib.h>
38 #include "lsp.h"
39 #include "os.h"
40 #include "misc.h"
41 #include "lookup.h"
42 #include "scales.h"
43 
44 /* three possible LSP to f curve functions; the exact computation
45    (float), a lookup based float implementation, and an integer
46    implementation.  The float lookup is likely the optimal choice on
47    any machine with an FPU.  The integer implementation is *not* fixed
48    point (due to the need for a large dynamic range and thus a
49    seperately tracked exponent) and thus much more complex than the
50    relatively simple float implementations. It's mostly for future
51    work on a fully fixed point implementation for processors like the
52    ARM family. */
53 
54 /* undefine both for the 'old' but more precise implementation */
55 #define   FLOAT_LOOKUP
56 #undef    INT_LOOKUP
57 
58 #ifdef FLOAT_LOOKUP
59 #include "lookup.c" /* catch this in the build system; we #include for
60                        compilers (like gcc) that can't inline across
61                        modules */
62 
63 /* side effect: changes *lsp to cosines of lsp */
vorbis_lsp_to_curve(float * curve,int * map,int n,int ln,float * lsp,int m,float amp,float ampoffset)64 void vorbis_lsp_to_curve(float *curve,int *map,int n,int ln,float *lsp,int m,
65 			    float amp,float ampoffset){
66   int i;
67   float wdel=M_PI/ln;
68   vorbis_fpu_control fpu;
69 
70   vorbis_fpu_setround(&fpu);
71   for(i=0;i<m;i++)lsp[i]=vorbis_coslook(lsp[i]);
72 
73   i=0;
74   while(i<n){
75     int k=map[i];
76     int qexp;
77     float p=.7071067812f;
78     float q=.7071067812f;
79     float w=vorbis_coslook(wdel*k);
80     float *ftmp=lsp;
81     int c=m>>1;
82 
83     do{
84       q*=ftmp[0]-w;
85       p*=ftmp[1]-w;
86       ftmp+=2;
87     }while(--c);
88 
89     if(m&1){
90       /* odd order filter; slightly assymetric */
91       /* the last coefficient */
92       q*=ftmp[0]-w;
93       q*=q;
94       p*=p*(1.f-w*w);
95     }else{
96       /* even order filter; still symmetric */
97       q*=q*(1.f+w);
98       p*=p*(1.f-w);
99     }
100 
101     q=frexp(p+q,&qexp);
102     q=vorbis_fromdBlook(amp*
103 			vorbis_invsqlook(q)*
104 			vorbis_invsq2explook(qexp+m)-
105 			ampoffset);
106 
107     do{
108       curve[i++]*=q;
109     }while(map[i]==k);
110   }
111   vorbis_fpu_restore(fpu);
112 }
113 
114 #else
115 
116 #ifdef INT_LOOKUP
117 #include "lookup.c" /* catch this in the build system; we #include for
118                        compilers (like gcc) that can't inline across
119                        modules */
120 
121 static int MLOOP_1[64]={
122    0,10,11,11, 12,12,12,12, 13,13,13,13, 13,13,13,13,
123   14,14,14,14, 14,14,14,14, 14,14,14,14, 14,14,14,14,
124   15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
125   15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
126 };
127 
128 static int MLOOP_2[64]={
129   0,4,5,5, 6,6,6,6, 7,7,7,7, 7,7,7,7,
130   8,8,8,8, 8,8,8,8, 8,8,8,8, 8,8,8,8,
131   9,9,9,9, 9,9,9,9, 9,9,9,9, 9,9,9,9,
132   9,9,9,9, 9,9,9,9, 9,9,9,9, 9,9,9,9,
133 };
134 
135 static int MLOOP_3[8]={0,1,2,2,3,3,3,3};
136 
137 
138 /* side effect: changes *lsp to cosines of lsp */
vorbis_lsp_to_curve(float * curve,int * map,int n,int ln,float * lsp,int m,float amp,float ampoffset)139 void vorbis_lsp_to_curve(float *curve,int *map,int n,int ln,float *lsp,int m,
140 			    float amp,float ampoffset){
141 
142   /* 0 <= m < 256 */
143 
144   /* set up for using all int later */
145   int i;
146   int ampoffseti=rint(ampoffset*4096.f);
147   int ampi=rint(amp*16.f);
148   long *ilsp=alloca(m*sizeof(*ilsp));
149   for(i=0;i<m;i++)ilsp[i]=vorbis_coslook_i(lsp[i]/M_PI*65536.f+.5f);
150 
151   i=0;
152   while(i<n){
153     int j,k=map[i];
154     unsigned long pi=46341; /* 2**-.5 in 0.16 */
155     unsigned long qi=46341;
156     int qexp=0,shift;
157     long wi=vorbis_coslook_i(k*65536/ln);
158 
159     qi*=labs(ilsp[0]-wi);
160     pi*=labs(ilsp[1]-wi);
161 
162     for(j=3;j<m;j+=2){
163       if(!(shift=MLOOP_1[(pi|qi)>>25]))
164 	if(!(shift=MLOOP_2[(pi|qi)>>19]))
165 	  shift=MLOOP_3[(pi|qi)>>16];
166       qi=(qi>>shift)*labs(ilsp[j-1]-wi);
167       pi=(pi>>shift)*labs(ilsp[j]-wi);
168       qexp+=shift;
169     }
170     if(!(shift=MLOOP_1[(pi|qi)>>25]))
171       if(!(shift=MLOOP_2[(pi|qi)>>19]))
172 	shift=MLOOP_3[(pi|qi)>>16];
173 
174     /* pi,qi normalized collectively, both tracked using qexp */
175 
176     if(m&1){
177       /* odd order filter; slightly assymetric */
178       /* the last coefficient */
179       qi=(qi>>shift)*labs(ilsp[j-1]-wi);
180       pi=(pi>>shift)<<14;
181       qexp+=shift;
182 
183       if(!(shift=MLOOP_1[(pi|qi)>>25]))
184 	if(!(shift=MLOOP_2[(pi|qi)>>19]))
185 	  shift=MLOOP_3[(pi|qi)>>16];
186 
187       pi>>=shift;
188       qi>>=shift;
189       qexp+=shift-14*((m+1)>>1);
190 
191       pi=((pi*pi)>>16);
192       qi=((qi*qi)>>16);
193       qexp=qexp*2+m;
194 
195       pi*=(1<<14)-((wi*wi)>>14);
196       qi+=pi>>14;
197 
198     }else{
199       /* even order filter; still symmetric */
200 
201       /* p*=p(1-w), q*=q(1+w), let normalization drift because it isn't
202 	 worth tracking step by step */
203 
204       pi>>=shift;
205       qi>>=shift;
206       qexp+=shift-7*m;
207 
208       pi=((pi*pi)>>16);
209       qi=((qi*qi)>>16);
210       qexp=qexp*2+m;
211 
212       pi*=(1<<14)-wi;
213       qi*=(1<<14)+wi;
214       qi=(qi+pi)>>14;
215 
216     }
217 
218 
219     /* we've let the normalization drift because it wasn't important;
220        however, for the lookup, things must be normalized again.  We
221        need at most one right shift or a number of left shifts */
222 
223     if(qi&0xffff0000){ /* checks for 1.xxxxxxxxxxxxxxxx */
224       qi>>=1; qexp++;
225     }else
226       while(qi && !(qi&0x8000)){ /* checks for 0.0xxxxxxxxxxxxxxx or less*/
227 	qi<<=1; qexp--;
228       }
229 
230     amp=vorbis_fromdBlook_i(ampi*                     /*  n.4         */
231 			    vorbis_invsqlook_i(qi,qexp)-
232 			                              /*  m.8, m+n<=8 */
233 			    ampoffseti);              /*  8.12[0]     */
234 
235     curve[i]*=amp;
236     while(map[++i]==k)curve[i]*=amp;
237   }
238 }
239 
240 #else
241 
242 /* old, nonoptimized but simple version for any poor sap who needs to
243    figure out what the hell this code does, or wants the other
244    fraction of a dB precision */
245 
246 /* side effect: changes *lsp to cosines of lsp */
vorbis_lsp_to_curve(float * curve,int * map,int n,int ln,float * lsp,int m,float amp,float ampoffset)247 void vorbis_lsp_to_curve(float *curve,int *map,int n,int ln,float *lsp,int m,
248 			    float amp,float ampoffset){
249   int i;
250   float wdel=M_PI/ln;
251   for(i=0;i<m;i++)lsp[i]=2.f*cos(lsp[i]);
252 
253   i=0;
254   while(i<n){
255     int j,k=map[i];
256     float p=.5f;
257     float q=.5f;
258     float w=2.f*cos(wdel*k);
259     for(j=1;j<m;j+=2){
260       q *= w-lsp[j-1];
261       p *= w-lsp[j];
262     }
263     if(j==m){
264       /* odd order filter; slightly assymetric */
265       /* the last coefficient */
266       q*=w-lsp[j-1];
267       p*=p*(4.f-w*w);
268       q*=q;
269     }else{
270       /* even order filter; still symmetric */
271       p*=p*(2.f-w);
272       q*=q*(2.f+w);
273     }
274 
275     q=fromdB(amp/sqrt(p+q)-ampoffset);
276 
277     curve[i]*=q;
278     while(map[++i]==k)curve[i]*=q;
279   }
280 }
281 
282 #endif
283 #endif
284 
cheby(float * g,int ord)285 static void cheby(float *g, int ord) {
286   int i, j;
287 
288   g[0] *= .5f;
289   for(i=2; i<= ord; i++) {
290     for(j=ord; j >= i; j--) {
291       g[j-2] -= g[j];
292       g[j] += g[j];
293     }
294   }
295 }
296 
comp(const void * a,const void * b)297 static int comp(const void *a,const void *b){
298   return (*(float *)a<*(float *)b)-(*(float *)a>*(float *)b);
299 }
300 
301 /* Newton-Raphson-Maehly actually functioned as a decent root finder,
302    but there are root sets for which it gets into limit cycles
303    (exacerbated by zero suppression) and fails.  We can't afford to
304    fail, even if the failure is 1 in 100,000,000, so we now use
305    Laguerre and later polish with Newton-Raphson (which can then
306    afford to fail) */
307 
308 #define EPSILON 10e-7
Laguerre_With_Deflation(float * a,int ord,float * r)309 static int Laguerre_With_Deflation(float *a,int ord,float *r){
310   int i,m;
311   double lastdelta=0.f;
312   double *defl=alloca(sizeof(*defl)*(ord+1));
313   for(i=0;i<=ord;i++)defl[i]=a[i];
314 
315   for(m=ord;m>0;m--){
316     double new=0.f,delta;
317 
318     /* iterate a root */
319     while(1){
320       double p=defl[m],pp=0.f,ppp=0.f,denom;
321 
322       /* eval the polynomial and its first two derivatives */
323       for(i=m;i>0;i--){
324 	ppp = new*ppp + pp;
325 	pp  = new*pp  + p;
326 	p   = new*p   + defl[i-1];
327       }
328 
329       /* Laguerre's method */
330       denom=(m-1) * ((m-1)*pp*pp - m*p*ppp);
331       if(denom<0)
332 	return(-1);  /* complex root!  The LPC generator handed us a bad filter */
333 
334       if(pp>0){
335 	denom = pp + sqrt(denom);
336 	if(denom<EPSILON)denom=EPSILON;
337       }else{
338 	denom = pp - sqrt(denom);
339 	if(denom>-(EPSILON))denom=-(EPSILON);
340       }
341 
342       delta  = m*p/denom;
343       new   -= delta;
344 
345       if(delta<0.f)delta*=-1;
346 
347       if(fabs(delta/new)<10e-12)break;
348       lastdelta=delta;
349     }
350 
351     r[m-1]=new;
352 
353     /* forward deflation */
354 
355     for(i=m;i>0;i--)
356       defl[i-1]+=new*defl[i];
357     defl++;
358 
359   }
360   return(0);
361 }
362 
363 
364 /* for spit-and-polish only */
Newton_Raphson(float * a,int ord,float * r)365 static int Newton_Raphson(float *a,int ord,float *r){
366   int i, k, count=0;
367   double error=1.f;
368   double *root=alloca(ord*sizeof(*root));
369 
370   for(i=0; i<ord;i++) root[i] = r[i];
371 
372   while(error>1e-20){
373     error=0;
374 
375     for(i=0; i<ord; i++) { /* Update each point. */
376       double pp=0.,delta;
377       double rooti=root[i];
378       double p=a[ord];
379       for(k=ord-1; k>= 0; k--) {
380 
381 	pp= pp* rooti + p;
382 	p = p * rooti + a[k];
383       }
384 
385       delta = p/pp;
386       root[i] -= delta;
387       error+= delta*delta;
388     }
389 
390     if(count>40)return(-1);
391 
392     count++;
393   }
394 
395   /* Replaced the original bubble sort with a real sort.  With your
396      help, we can eliminate the bubble sort in our lifetime. --Monty */
397 
398   for(i=0; i<ord;i++) r[i] = root[i];
399   return(0);
400 }
401 
402 
403 /* Convert lpc coefficients to lsp coefficients */
vorbis_lpc_to_lsp(float * lpc,float * lsp,int m)404 int vorbis_lpc_to_lsp(float *lpc,float *lsp,int m){
405   int order2=(m+1)>>1;
406   int g1_order,g2_order;
407   float *g1=alloca(sizeof(*g1)*(order2+1));
408   float *g2=alloca(sizeof(*g2)*(order2+1));
409   float *g1r=alloca(sizeof(*g1r)*(order2+1));
410   float *g2r=alloca(sizeof(*g2r)*(order2+1));
411   int i;
412 
413   /* even and odd are slightly different base cases */
414   g1_order=(m+1)>>1;
415   g2_order=(m)  >>1;
416 
417   /* Compute the lengths of the x polynomials. */
418   /* Compute the first half of K & R F1 & F2 polynomials. */
419   /* Compute half of the symmetric and antisymmetric polynomials. */
420   /* Remove the roots at +1 and -1. */
421 
422   g1[g1_order] = 1.f;
423   for(i=1;i<=g1_order;i++) g1[g1_order-i] = lpc[i-1]+lpc[m-i];
424   g2[g2_order] = 1.f;
425   for(i=1;i<=g2_order;i++) g2[g2_order-i] = lpc[i-1]-lpc[m-i];
426 
427   if(g1_order>g2_order){
428     for(i=2; i<=g2_order;i++) g2[g2_order-i] += g2[g2_order-i+2];
429   }else{
430     for(i=1; i<=g1_order;i++) g1[g1_order-i] -= g1[g1_order-i+1];
431     for(i=1; i<=g2_order;i++) g2[g2_order-i] += g2[g2_order-i+1];
432   }
433 
434   /* Convert into polynomials in cos(alpha) */
435   cheby(g1,g1_order);
436   cheby(g2,g2_order);
437 
438   /* Find the roots of the 2 even polynomials.*/
439   if(Laguerre_With_Deflation(g1,g1_order,g1r) ||
440      Laguerre_With_Deflation(g2,g2_order,g2r))
441     return(-1);
442 
443   Newton_Raphson(g1,g1_order,g1r); /* if it fails, it leaves g1r alone */
444   Newton_Raphson(g2,g2_order,g2r); /* if it fails, it leaves g2r alone */
445 
446   qsort(g1r,g1_order,sizeof(*g1r),comp);
447   qsort(g2r,g2_order,sizeof(*g2r),comp);
448 
449   for(i=0;i<g1_order;i++)
450     lsp[i*2] = acos(g1r[i]);
451 
452   for(i=0;i<g2_order;i++)
453     lsp[i*2+1] = acos(g2r[i]);
454   return(0);
455 }
456