FFmpeg: libswresample/resample.c Source File

00001 /*
00002  * audio resampling
00003  * Copyright (c) 2004 Michael Niedermayer <michaelni@gmx.at>
00004  *
00005  * This file is part of FFmpeg.
00006  *
00007  * FFmpeg is free software; you can redistribute it and/or
00008  * modify it under the terms of the GNU Lesser General Public
00009  * License as published by the Free Software Foundation; either
00010  * version 2.1 of the License, or (at your option) any later version.
00011  *
00012  * FFmpeg is distributed in the hope that it will be useful,
00013  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00014  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00015  * Lesser General Public License for more details.
00016  *
00017  * You should have received a copy of the GNU Lesser General Public
00018  * License along with FFmpeg; if not, write to the Free Software
00019  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00020  */
00021 
00028 #include "libavutil/log.h"
00029 #include "swresample_internal.h"
00030 
00031 #ifndef CONFIG_RESAMPLE_HP
00032 #define FILTER_SHIFT 15
00033 
00034 #define FELEM int16_t
00035 #define FELEM2 int32_t
00036 #define FELEML int64_t
00037 #define FELEM_MAX INT16_MAX
00038 #define FELEM_MIN INT16_MIN
00039 #define WINDOW_TYPE 9
00040 #elif !defined(CONFIG_RESAMPLE_AUDIOPHILE_KIDDY_MODE)
00041 #define FILTER_SHIFT 30
00042 
00043 #define FELEM int32_t
00044 #define FELEM2 int64_t
00045 #define FELEML int64_t
00046 #define FELEM_MAX INT32_MAX
00047 #define FELEM_MIN INT32_MIN
00048 #define WINDOW_TYPE 12
00049 #else
00050 #define FILTER_SHIFT 0
00051 
00052 #define FELEM double
00053 #define FELEM2 double
00054 #define FELEML double
00055 #define WINDOW_TYPE 24
00056 #endif
00057 
00058 
00059 typedef struct ResampleContext {
00060     const AVClass *av_class;
00061     FELEM *filter_bank;
00062     int filter_length;
00063     int ideal_dst_incr;
00064     int dst_incr;
00065     int index;
00066     int frac;
00067     int src_incr;
00068     int compensation_distance;
00069     int phase_shift;
00070     int phase_mask;
00071     int linear;
00072     double factor;
00073 } ResampleContext;
00074 
00078 static double bessel(double x){
00079     double v=1;
00080     double lastv=0;
00081     double t=1;
00082     int i;
00083     static const double inv[100]={
00084  1.0/( 1* 1), 1.0/( 2* 2), 1.0/( 3* 3), 1.0/( 4* 4), 1.0/( 5* 5), 1.0/( 6* 6), 1.0/( 7* 7), 1.0/( 8* 8), 1.0/( 9* 9), 1.0/(10*10),
00085  1.0/(11*11), 1.0/(12*12), 1.0/(13*13), 1.0/(14*14), 1.0/(15*15), 1.0/(16*16), 1.0/(17*17), 1.0/(18*18), 1.0/(19*19), 1.0/(20*20),
00086  1.0/(21*21), 1.0/(22*22), 1.0/(23*23), 1.0/(24*24), 1.0/(25*25), 1.0/(26*26), 1.0/(27*27), 1.0/(28*28), 1.0/(29*29), 1.0/(30*30),
00087  1.0/(31*31), 1.0/(32*32), 1.0/(33*33), 1.0/(34*34), 1.0/(35*35), 1.0/(36*36), 1.0/(37*37), 1.0/(38*38), 1.0/(39*39), 1.0/(40*40),
00088  1.0/(41*41), 1.0/(42*42), 1.0/(43*43), 1.0/(44*44), 1.0/(45*45), 1.0/(46*46), 1.0/(47*47), 1.0/(48*48), 1.0/(49*49), 1.0/(50*50),
00089  1.0/(51*51), 1.0/(52*52), 1.0/(53*53), 1.0/(54*54), 1.0/(55*55), 1.0/(56*56), 1.0/(57*57), 1.0/(58*58), 1.0/(59*59), 1.0/(60*60),
00090  1.0/(61*61), 1.0/(62*62), 1.0/(63*63), 1.0/(64*64), 1.0/(65*65), 1.0/(66*66), 1.0/(67*67), 1.0/(68*68), 1.0/(69*69), 1.0/(70*70),
00091  1.0/(71*71), 1.0/(72*72), 1.0/(73*73), 1.0/(74*74), 1.0/(75*75), 1.0/(76*76), 1.0/(77*77), 1.0/(78*78), 1.0/(79*79), 1.0/(80*80),
00092  1.0/(81*81), 1.0/(82*82), 1.0/(83*83), 1.0/(84*84), 1.0/(85*85), 1.0/(86*86), 1.0/(87*87), 1.0/(88*88), 1.0/(89*89), 1.0/(90*90),
00093  1.0/(91*91), 1.0/(92*92), 1.0/(93*93), 1.0/(94*94), 1.0/(95*95), 1.0/(96*96), 1.0/(97*97), 1.0/(98*98), 1.0/(99*99), 1.0/(10000)
00094     };
00095 
00096     x= x*x/4;
00097     for(i=0; v != lastv; i++){
00098         lastv=v;
00099         t *= x*inv[i];
00100         v += t;
00101     }
00102     return v;
00103 }
00104 
00112 static int build_filter(FELEM *filter, double factor, int tap_count, int phase_count, int scale, int type){
00113     int ph, i;
00114     double x, y, w;
00115     double *tab = av_malloc(tap_count * sizeof(*tab));
00116     const int center= (tap_count-1)/2;
00117 
00118     if (!tab)
00119         return AVERROR(ENOMEM);
00120 
00121     /* if upsampling, only need to interpolate, no filter */
00122     if (factor > 1.0)
00123         factor = 1.0;
00124 
00125     for(ph=0;ph<phase_count;ph++) {
00126         double norm = 0;
00127         for(i=0;i<tap_count;i++) {
00128             x = M_PI * ((double)(i - center) - (double)ph / phase_count) * factor;
00129             if (x == 0) y = 1.0;
00130             else        y = sin(x) / x;
00131             switch(type){
00132             case 0:{
00133                 const float d= -0.5; //first order derivative = -0.5
00134                 x = fabs(((double)(i - center) - (double)ph / phase_count) * factor);
00135                 if(x<1.0) y= 1 - 3*x*x + 2*x*x*x + d*(            -x*x + x*x*x);
00136                 else      y=                       d*(-4 + 8*x - 5*x*x + x*x*x);
00137                 break;}
00138             case 1:
00139                 w = 2.0*x / (factor*tap_count) + M_PI;
00140                 y *= 0.3635819 - 0.4891775 * cos(w) + 0.1365995 * cos(2*w) - 0.0106411 * cos(3*w);
00141                 break;
00142             default:
00143                 w = 2.0*x / (factor*tap_count*M_PI);
00144                 y *= bessel(type*sqrt(FFMAX(1-w*w, 0)));
00145                 break;
00146             }
00147 
00148             tab[i] = y;
00149             norm += y;
00150         }
00151 
00152         /* normalize so that an uniform color remains the same */
00153         for(i=0;i<tap_count;i++) {
00154 #ifdef CONFIG_RESAMPLE_AUDIOPHILE_KIDDY_MODE
00155             filter[ph * tap_count + i] = tab[i] / norm;
00156 #else
00157             filter[ph * tap_count + i] = av_clip(lrintf(tab[i] * scale / norm), FELEM_MIN, FELEM_MAX);
00158 #endif
00159         }
00160     }
00161 #if 0
00162     {
00163 #define LEN 1024
00164         int j,k;
00165         double sine[LEN + tap_count];
00166         double filtered[LEN];
00167         double maxff=-2, minff=2, maxsf=-2, minsf=2;
00168         for(i=0; i<LEN; i++){
00169             double ss=0, sf=0, ff=0;
00170             for(j=0; j<LEN+tap_count; j++)
00171                 sine[j]= cos(i*j*M_PI/LEN);
00172             for(j=0; j<LEN; j++){
00173                 double sum=0;
00174                 ph=0;
00175                 for(k=0; k<tap_count; k++)
00176                     sum += filter[ph * tap_count + k] * sine[k+j];
00177                 filtered[j]= sum / (1<<FILTER_SHIFT);
00178                 ss+= sine[j + center] * sine[j + center];
00179                 ff+= filtered[j] * filtered[j];
00180                 sf+= sine[j + center] * filtered[j];
00181             }
00182             ss= sqrt(2*ss/LEN);
00183             ff= sqrt(2*ff/LEN);
00184             sf= 2*sf/LEN;
00185             maxff= FFMAX(maxff, ff);
00186             minff= FFMIN(minff, ff);
00187             maxsf= FFMAX(maxsf, sf);
00188             minsf= FFMIN(minsf, sf);
00189             if(i%11==0){
00190                 av_log(NULL, AV_LOG_ERROR, "i:%4d ss:%f ff:%13.6e-%13.6e sf:%13.6e-%13.6e\n", i, ss, maxff, minff, maxsf, minsf);
00191                 minff=minsf= 2;
00192                 maxff=maxsf= -2;
00193             }
00194         }
00195     }
00196 #endif
00197 
00198     av_free(tab);
00199     return 0;
00200 }
00201 
00202 ResampleContext *swri_resample_init(ResampleContext *c, int out_rate, int in_rate, int filter_size, int phase_shift, int linear, double cutoff){
00203     double factor= FFMIN(out_rate * cutoff / in_rate, 1.0);
00204     int phase_count= 1<<phase_shift;
00205 
00206     if (!c || c->phase_shift != phase_shift || c->linear!=linear || c->factor != factor
00207            || c->filter_length != FFMAX((int)ceil(filter_size/factor), 1)) {
00208         c = av_mallocz(sizeof(*c));
00209         if (!c)
00210             return NULL;
00211 
00212         c->phase_shift   = phase_shift;
00213         c->phase_mask    = phase_count - 1;
00214         c->linear        = linear;
00215         c->factor        = factor;
00216         c->filter_length = FFMAX((int)ceil(filter_size/factor), 1);
00217         c->filter_bank   = av_mallocz(c->filter_length*(phase_count+1)*sizeof(FELEM));
00218         if (!c->filter_bank)
00219             goto error;
00220         if (build_filter(c->filter_bank, factor, c->filter_length, phase_count, 1<<FILTER_SHIFT, WINDOW_TYPE))
00221             goto error;
00222         memcpy(&c->filter_bank[c->filter_length*phase_count+1], c->filter_bank, (c->filter_length-1)*sizeof(FELEM));
00223         c->filter_bank[c->filter_length*phase_count]= c->filter_bank[c->filter_length - 1];
00224     }
00225 
00226     c->compensation_distance= 0;
00227     if(!av_reduce(&c->src_incr, &c->dst_incr, out_rate, in_rate * (int64_t)phase_count, INT32_MAX/2))
00228         goto error;
00229     c->ideal_dst_incr= c->dst_incr;
00230 
00231     c->index= -phase_count*((c->filter_length-1)/2);
00232     c->frac= 0;
00233 
00234     return c;
00235 error:
00236     av_free(c->filter_bank);
00237     av_free(c);
00238     return NULL;
00239 }
00240 
00241 void swri_resample_free(ResampleContext **c){
00242     if(!*c)
00243         return;
00244     av_freep(&(*c)->filter_bank);
00245     av_freep(c);
00246 }
00247 
00248 int swr_set_compensation(struct SwrContext *s, int sample_delta, int compensation_distance){
00249     ResampleContext *c;
00250     int ret;
00251 
00252     if (!s || compensation_distance < 0)
00253         return AVERROR(EINVAL);
00254     if (!compensation_distance && sample_delta)
00255         return AVERROR(EINVAL);
00256     if (!s->resample) {
00257         s->flags |= SWR_FLAG_RESAMPLE;
00258         ret = swr_init(s);
00259         if (ret < 0)
00260             return ret;
00261     }
00262     c= s->resample;
00263     c->compensation_distance= compensation_distance;
00264     if (compensation_distance)
00265         c->dst_incr = c->ideal_dst_incr - c->ideal_dst_incr * (int64_t)sample_delta / compensation_distance;
00266     else
00267         c->dst_incr = c->ideal_dst_incr;
00268     return 0;
00269 }
00270 
00271 int swri_resample(ResampleContext *c, int16_t *dst, const int16_t *src, int *consumed, int src_size, int dst_size, int update_ctx){
00272     int dst_index, i;
00273     int index= c->index;
00274     int frac= c->frac;
00275     int dst_incr_frac= c->dst_incr % c->src_incr;
00276     int dst_incr=      c->dst_incr / c->src_incr;
00277     int compensation_distance= c->compensation_distance;
00278 
00279     if(compensation_distance == 0 && c->filter_length == 1 && c->phase_shift==0){
00280         int64_t index2= ((int64_t)index)<<32;
00281         int64_t incr= (1LL<<32) * c->dst_incr / c->src_incr;
00282         dst_size= FFMIN(dst_size, (src_size-1-index) * (int64_t)c->src_incr / c->dst_incr);
00283 
00284         for(dst_index=0; dst_index < dst_size; dst_index++){
00285             dst[dst_index] = src[index2>>32];
00286             index2 += incr;
00287         }
00288         index += dst_index * dst_incr;
00289         index += (frac + dst_index * (int64_t)dst_incr_frac) / c->src_incr;
00290         frac   = (frac + dst_index * (int64_t)dst_incr_frac) % c->src_incr;
00291     }else{
00292         for(dst_index=0; dst_index < dst_size; dst_index++){
00293             FELEM *filter= c->filter_bank + c->filter_length*(index & c->phase_mask);
00294             int sample_index= index >> c->phase_shift;
00295             FELEM2 val=0;
00296 
00297             if(sample_index < 0){
00298                 for(i=0; i<c->filter_length; i++)
00299                     val += src[FFABS(sample_index + i) % src_size] * filter[i];
00300             }else if(sample_index + c->filter_length > src_size){
00301                 break;
00302             }else if(c->linear){
00303                 FELEM2 v2=0;
00304                 for(i=0; i<c->filter_length; i++){
00305                     val += src[sample_index + i] * (FELEM2)filter[i];
00306                     v2  += src[sample_index + i] * (FELEM2)filter[i + c->filter_length];
00307                 }
00308                 val+=(v2-val)*(FELEML)frac / c->src_incr;
00309             }else{
00310                 for(i=0; i<c->filter_length; i++){
00311                     val += src[sample_index + i] * (FELEM2)filter[i];
00312                 }
00313             }
00314 
00315 #ifdef CONFIG_RESAMPLE_AUDIOPHILE_KIDDY_MODE
00316             dst[dst_index] = av_clip_int16(lrintf(val));
00317 #else
00318             val = (val + (1<<(FILTER_SHIFT-1)))>>FILTER_SHIFT;
00319             dst[dst_index] = (unsigned)(val + 32768) > 65535 ? (val>>31) ^ 32767 : val;
00320 #endif
00321 
00322             frac += dst_incr_frac;
00323             index += dst_incr;
00324             if(frac >= c->src_incr){
00325                 frac -= c->src_incr;
00326                 index++;
00327             }
00328 
00329             if(dst_index + 1 == compensation_distance){
00330                 compensation_distance= 0;
00331                 dst_incr_frac= c->ideal_dst_incr % c->src_incr;
00332                 dst_incr=      c->ideal_dst_incr / c->src_incr;
00333             }
00334         }
00335     }
00336     *consumed= FFMAX(index, 0) >> c->phase_shift;
00337     if(index>=0) index &= c->phase_mask;
00338 
00339     if(compensation_distance){
00340         compensation_distance -= dst_index;
00341         assert(compensation_distance > 0);
00342     }
00343     if(update_ctx){
00344         c->frac= frac;
00345         c->index= index;
00346         c->dst_incr= dst_incr_frac + c->src_incr*dst_incr;
00347         c->compensation_distance= compensation_distance;
00348     }
00349 #if 0
00350     if(update_ctx && !c->compensation_distance){
00351 #undef rand
00352         av_resample_compensate(c, rand() % (8000*2) - 8000, 8000*2);
00353 av_log(NULL, AV_LOG_DEBUG, "%d %d %d\n", c->dst_incr, c->ideal_dst_incr, c->compensation_distance);
00354     }
00355 #endif
00356 
00357     return dst_index;
00358 }
00359 
00360 int swri_multiple_resample(ResampleContext *c, AudioData *dst, int dst_size, AudioData *src, int src_size, int *consumed){
00361     int i, ret= -1;
00362 
00363     for(i=0; i<dst->ch_count; i++){
00364         ret= swri_resample(c, (int16_t*)dst->ch[i], (const int16_t*)src->ch[i], consumed, src_size, dst_size, i+1==dst->ch_count);
00365     }
00366 
00367     return ret;
00368 }