libswresample/swresample.c
Go to the documentation of this file.
00001 /*
00002  * Copyright (C) 2011 Michael Niedermayer (michaelni@gmx.at)
00003  *
00004  * This file is part of libswresample
00005  *
00006  * libswresample is free software; you can redistribute it and/or
00007  * modify it under the terms of the GNU Lesser General Public
00008  * License as published by the Free Software Foundation; either
00009  * version 2.1 of the License, or (at your option) any later version.
00010  *
00011  * libswresample is distributed in the hope that it will be useful,
00012  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00013  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00014  * Lesser General Public License for more details.
00015  *
00016  * You should have received a copy of the GNU Lesser General Public
00017  * License along with libswresample; if not, write to the Free Software
00018  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00019  */
00020 
00021 #include "libavutil/opt.h"
00022 #include "swresample_internal.h"
00023 #include "audioconvert.h"
00024 #include "libavutil/avassert.h"
00025 #include "libavutil/audioconvert.h"
00026 
00027 #define  C30DB  M_SQRT2
00028 #define  C15DB  1.189207115
00029 #define C__0DB  1.0
00030 #define C_15DB  0.840896415
00031 #define C_30DB  M_SQRT1_2
00032 #define C_45DB  0.594603558
00033 #define C_60DB  0.5
00034 
00035 
00036 //TODO split options array out?
00037 #define OFFSET(x) offsetof(SwrContext,x)
00038 static const AVOption options[]={
00039 {"ich",  "input channel count", OFFSET( in.ch_count   ), AV_OPT_TYPE_INT, {.dbl=2}, 0, SWR_CH_MAX, 0},
00040 {"och", "output channel count", OFFSET(out.ch_count   ), AV_OPT_TYPE_INT, {.dbl=2}, 0, SWR_CH_MAX, 0},
00041 {"uch",   "used channel count", OFFSET(used_ch_count  ), AV_OPT_TYPE_INT, {.dbl=0}, 0, SWR_CH_MAX, 0},
00042 {"isr",  "input sample rate"  , OFFSET( in_sample_rate), AV_OPT_TYPE_INT, {.dbl=48000}, 1, INT_MAX, 0},
00043 {"osr", "output sample rate"  , OFFSET(out_sample_rate), AV_OPT_TYPE_INT, {.dbl=48000}, 1, INT_MAX, 0},
00044 //{"ip" ,  "input planar"       , OFFSET( in.planar     ), AV_OPT_TYPE_INT, {.dbl=0},    0,       1, 0},
00045 //{"op" , "output planar"       , OFFSET(out.planar     ), AV_OPT_TYPE_INT, {.dbl=0},    0,       1, 0},
00046 {"isf",  "input sample format", OFFSET( in_sample_fmt ), AV_OPT_TYPE_INT, {.dbl=AV_SAMPLE_FMT_S16}, 0, AV_SAMPLE_FMT_NB-1+256, 0},
00047 {"osf", "output sample format", OFFSET(out_sample_fmt ), AV_OPT_TYPE_INT, {.dbl=AV_SAMPLE_FMT_S16}, 0, AV_SAMPLE_FMT_NB-1+256, 0},
00048 {"tsf", "internal sample format", OFFSET(int_sample_fmt ), AV_OPT_TYPE_INT, {.dbl=AV_SAMPLE_FMT_NONE}, -1, AV_SAMPLE_FMT_FLT, 0},
00049 {"icl",  "input channel layout" , OFFSET( in_ch_layout), AV_OPT_TYPE_INT64, {.dbl=0}, 0, INT64_MAX, 0, "channel_layout"},
00050 {"ocl",  "output channel layout", OFFSET(out_ch_layout), AV_OPT_TYPE_INT64, {.dbl=0}, 0, INT64_MAX, 0, "channel_layout"},
00051 {"clev", "center mix level"     , OFFSET(clev)         , AV_OPT_TYPE_FLOAT, {.dbl=C_30DB}, 0, 4, 0},
00052 {"slev", "sourround mix level"  , OFFSET(slev)         , AV_OPT_TYPE_FLOAT, {.dbl=C_30DB}, 0, 4, 0},
00053 {"rmvol", "rematrix volume"     , OFFSET(rematrix_volume), AV_OPT_TYPE_FLOAT, {.dbl=1.0}, -1000, 1000, 0},
00054 {"flags", NULL                  , OFFSET(flags)        , AV_OPT_TYPE_FLAGS, {.dbl=0}, 0,  UINT_MAX, 0, "flags"},
00055 {"res", "force resampling", 0, AV_OPT_TYPE_CONST, {.dbl=SWR_FLAG_RESAMPLE}, INT_MIN, INT_MAX, 0, "flags"},
00056 
00057 {0}
00058 };
00059 
00060 static const char* context_to_name(void* ptr) {
00061     return "SWR";
00062 }
00063 
00064 static const AVClass av_class = {
00065     .class_name                = "SwrContext",
00066     .item_name                 = context_to_name,
00067     .option                    = options,
00068     .version                   = LIBAVUTIL_VERSION_INT,
00069     .log_level_offset_offset   = OFFSET(log_level_offset),
00070     .parent_log_context_offset = OFFSET(log_ctx),
00071 };
00072 
00073 unsigned swresample_version(void)
00074 {
00075     av_assert0(LIBSWRESAMPLE_VERSION_MICRO >= 100);
00076     return LIBSWRESAMPLE_VERSION_INT;
00077 }
00078 
00079 const char *swresample_configuration(void)
00080 {
00081     return FFMPEG_CONFIGURATION;
00082 }
00083 
00084 const char *swresample_license(void)
00085 {
00086 #define LICENSE_PREFIX "libswresample license: "
00087     return LICENSE_PREFIX FFMPEG_LICENSE + sizeof(LICENSE_PREFIX) - 1;
00088 }
00089 
00090 int swr_set_channel_mapping(struct SwrContext *s, const int *channel_map){
00091     if(!s || s->in_convert) // s needs to be allocated but not initialized
00092         return AVERROR(EINVAL);
00093     s->channel_map = channel_map;
00094     return 0;
00095 }
00096 
00097 struct SwrContext *swr_alloc(void){
00098     SwrContext *s= av_mallocz(sizeof(SwrContext));
00099     if(s){
00100         s->av_class= &av_class;
00101         av_opt_set_defaults(s);
00102     }
00103     return s;
00104 }
00105 
00106 struct SwrContext *swr_alloc_set_opts(struct SwrContext *s,
00107                                       int64_t out_ch_layout, enum AVSampleFormat out_sample_fmt, int out_sample_rate,
00108                                       int64_t  in_ch_layout, enum AVSampleFormat  in_sample_fmt, int  in_sample_rate,
00109                                       int log_offset, void *log_ctx){
00110     if(!s) s= swr_alloc();
00111     if(!s) return NULL;
00112 
00113     s->log_level_offset= log_offset;
00114     s->log_ctx= log_ctx;
00115 
00116     av_opt_set_int(s, "ocl", out_ch_layout,   0);
00117     av_opt_set_int(s, "osf", out_sample_fmt,  0);
00118     av_opt_set_int(s, "osr", out_sample_rate, 0);
00119     av_opt_set_int(s, "icl", in_ch_layout,    0);
00120     av_opt_set_int(s, "isf", in_sample_fmt,   0);
00121     av_opt_set_int(s, "isr", in_sample_rate,  0);
00122     av_opt_set_int(s, "tsf", AV_SAMPLE_FMT_S16, 0);
00123     av_opt_set_int(s, "ich", av_get_channel_layout_nb_channels(s-> in_ch_layout), 0);
00124     av_opt_set_int(s, "och", av_get_channel_layout_nb_channels(s->out_ch_layout), 0);
00125     av_opt_set_int(s, "uch", 0, 0);
00126     return s;
00127 }
00128 
00129 
00130 static void free_temp(AudioData *a){
00131     av_free(a->data);
00132     memset(a, 0, sizeof(*a));
00133 }
00134 
00135 void swr_free(SwrContext **ss){
00136     SwrContext *s= *ss;
00137     if(s){
00138         free_temp(&s->postin);
00139         free_temp(&s->midbuf);
00140         free_temp(&s->preout);
00141         free_temp(&s->in_buffer);
00142         swri_audio_convert_free(&s-> in_convert);
00143         swri_audio_convert_free(&s->out_convert);
00144         swri_audio_convert_free(&s->full_convert);
00145         swri_resample_free(&s->resample);
00146     }
00147 
00148     av_freep(ss);
00149 }
00150 
00151 int swr_init(struct SwrContext *s){
00152     s->in_buffer_index= 0;
00153     s->in_buffer_count= 0;
00154     s->resample_in_constraint= 0;
00155     free_temp(&s->postin);
00156     free_temp(&s->midbuf);
00157     free_temp(&s->preout);
00158     free_temp(&s->in_buffer);
00159     swri_audio_convert_free(&s-> in_convert);
00160     swri_audio_convert_free(&s->out_convert);
00161     swri_audio_convert_free(&s->full_convert);
00162 
00163     s-> in.planar= av_sample_fmt_is_planar(s-> in_sample_fmt);
00164     s->out.planar= av_sample_fmt_is_planar(s->out_sample_fmt);
00165     s-> in_sample_fmt= av_get_alt_sample_fmt(s-> in_sample_fmt, 0);
00166     s->out_sample_fmt= av_get_alt_sample_fmt(s->out_sample_fmt, 0);
00167 
00168     if(s-> in_sample_fmt >= AV_SAMPLE_FMT_NB){
00169         av_log(s, AV_LOG_ERROR, "Requested input sample format %d is invalid\n", s->in_sample_fmt);
00170         return AVERROR(EINVAL);
00171     }
00172     if(s->out_sample_fmt >= AV_SAMPLE_FMT_NB){
00173         av_log(s, AV_LOG_ERROR, "Requested output sample format %d is invalid\n", s->out_sample_fmt);
00174         return AVERROR(EINVAL);
00175     }
00176 
00177     if(   s->int_sample_fmt != AV_SAMPLE_FMT_S16
00178         &&s->int_sample_fmt != AV_SAMPLE_FMT_FLT){
00179         av_log(s, AV_LOG_ERROR, "Requested sample format %s is not supported internally, only float & S16 is supported\n", av_get_sample_fmt_name(s->int_sample_fmt));
00180         return AVERROR(EINVAL);
00181     }
00182 
00183     //FIXME should we allow/support using FLT on material that doesnt need it ?
00184     if(s->in_sample_fmt <= AV_SAMPLE_FMT_S16 || s->int_sample_fmt==AV_SAMPLE_FMT_S16){
00185         s->int_sample_fmt= AV_SAMPLE_FMT_S16;
00186     }else
00187         s->int_sample_fmt= AV_SAMPLE_FMT_FLT;
00188 
00189 
00190     if (s->out_sample_rate!=s->in_sample_rate || (s->flags & SWR_FLAG_RESAMPLE)){
00191         s->resample = swri_resample_init(s->resample, s->out_sample_rate, s->in_sample_rate, 16, 10, 0, 0.8);
00192     }else
00193         swri_resample_free(&s->resample);
00194     if(s->int_sample_fmt != AV_SAMPLE_FMT_S16 && s->resample){
00195         av_log(s, AV_LOG_ERROR, "Resampling only supported with internal s16 currently\n"); //FIXME
00196         return -1;
00197     }
00198 
00199     if(!s->used_ch_count)
00200         s->used_ch_count= s->in.ch_count;
00201 
00202     if(s->used_ch_count && s-> in_ch_layout && s->used_ch_count != av_get_channel_layout_nb_channels(s-> in_ch_layout)){
00203         av_log(s, AV_LOG_WARNING, "Input channel layout has a different number of channels than the number of used channels, ignoring layout\n");
00204         s-> in_ch_layout= 0;
00205     }
00206 
00207     if(!s-> in_ch_layout)
00208         s-> in_ch_layout= av_get_default_channel_layout(s->used_ch_count);
00209     if(!s->out_ch_layout)
00210         s->out_ch_layout= av_get_default_channel_layout(s->out.ch_count);
00211 
00212     s->rematrix= s->out_ch_layout  !=s->in_ch_layout || s->rematrix_volume!=1.0;
00213 
00214 #define RSC 1 //FIXME finetune
00215     if(!s-> in.ch_count)
00216         s-> in.ch_count= av_get_channel_layout_nb_channels(s-> in_ch_layout);
00217     if(!s->used_ch_count)
00218         s->used_ch_count= s->in.ch_count;
00219     if(!s->out.ch_count)
00220         s->out.ch_count= av_get_channel_layout_nb_channels(s->out_ch_layout);
00221 
00222     if(!s-> in.ch_count){
00223         av_assert0(!s->in_ch_layout);
00224         av_log(s, AV_LOG_ERROR, "Input channel count and layout are unset\n");
00225         return -1;
00226     }
00227 
00228 av_assert0(s->used_ch_count);
00229 av_assert0(s->out.ch_count);
00230     s->resample_first= RSC*s->out.ch_count/s->in.ch_count - RSC < s->out_sample_rate/(float)s-> in_sample_rate - 1.0;
00231 
00232     s-> in.bps= av_get_bytes_per_sample(s-> in_sample_fmt);
00233     s->int_bps= av_get_bytes_per_sample(s->int_sample_fmt);
00234     s->out.bps= av_get_bytes_per_sample(s->out_sample_fmt);
00235 
00236     if(!s->resample && !s->rematrix && !s->channel_map){
00237         s->full_convert = swri_audio_convert_alloc(s->out_sample_fmt,
00238                                                    s-> in_sample_fmt, s-> in.ch_count, NULL, 0);
00239         return 0;
00240     }
00241 
00242     s->in_convert = swri_audio_convert_alloc(s->int_sample_fmt,
00243                                              s-> in_sample_fmt, s->used_ch_count, s->channel_map, 0);
00244     s->out_convert= swri_audio_convert_alloc(s->out_sample_fmt,
00245                                              s->int_sample_fmt, s->out.ch_count, NULL, 0);
00246 
00247 
00248     s->postin= s->in;
00249     s->preout= s->out;
00250     s->midbuf= s->in;
00251     s->in_buffer= s->in;
00252     if(s->channel_map){
00253         s->postin.ch_count=
00254         s->midbuf.ch_count=
00255         s->in_buffer.ch_count= s->used_ch_count;
00256     }
00257     if(!s->resample_first){
00258         s->midbuf.ch_count= s->out.ch_count;
00259         s->in_buffer.ch_count = s->out.ch_count;
00260     }
00261 
00262     s->in_buffer.bps = s->postin.bps = s->midbuf.bps = s->preout.bps =  s->int_bps;
00263     s->in_buffer.planar = s->postin.planar = s->midbuf.planar = s->preout.planar =  1;
00264 
00265 
00266     if(s->rematrix)
00267         return swri_rematrix_init(s);
00268 
00269     return 0;
00270 }
00271 
00272 static int realloc_audio(AudioData *a, int count){
00273     int i, countb;
00274     AudioData old;
00275 
00276     if(a->count >= count)
00277         return 0;
00278 
00279     count*=2;
00280 
00281     countb= FFALIGN(count*a->bps, 32);
00282     old= *a;
00283 
00284     av_assert0(a->planar);
00285     av_assert0(a->bps);
00286     av_assert0(a->ch_count);
00287 
00288     a->data= av_malloc(countb*a->ch_count);
00289     if(!a->data)
00290         return AVERROR(ENOMEM);
00291     for(i=0; i<a->ch_count; i++){
00292         a->ch[i]= a->data + i*(a->planar ? countb : a->bps);
00293         if(a->planar) memcpy(a->ch[i], old.ch[i], a->count*a->bps);
00294     }
00295     av_free(old.data);
00296     a->count= count;
00297 
00298     return 1;
00299 }
00300 
00301 static void copy(AudioData *out, AudioData *in,
00302                  int count){
00303     av_assert0(out->planar == in->planar);
00304     av_assert0(out->bps == in->bps);
00305     av_assert0(out->ch_count == in->ch_count);
00306     if(out->planar){
00307         int ch;
00308         for(ch=0; ch<out->ch_count; ch++)
00309             memcpy(out->ch[ch], in->ch[ch], count*out->bps);
00310     }else
00311         memcpy(out->ch[0], in->ch[0], count*out->ch_count*out->bps);
00312 }
00313 
00314 static void fill_audiodata(AudioData *out, uint8_t *in_arg [SWR_CH_MAX]){
00315     int i;
00316     if(out->planar){
00317         for(i=0; i<out->ch_count; i++)
00318             out->ch[i]= in_arg[i];
00319     }else{
00320         for(i=0; i<out->ch_count; i++)
00321             out->ch[i]= in_arg[0] + i*out->bps;
00322     }
00323 }
00324 
00329 static void buf_set(AudioData *out, AudioData *in, int count){
00330     if(in->planar){
00331         int ch;
00332         for(ch=0; ch<out->ch_count; ch++)
00333             out->ch[ch]= in->ch[ch] + count*out->bps;
00334     }else
00335         out->ch[0]= in->ch[0] + count*out->ch_count*out->bps;
00336 }
00337 
00342 static int resample(SwrContext *s, AudioData *out_param, int out_count,
00343                              const AudioData * in_param, int in_count){
00344     AudioData in, out, tmp;
00345     int ret_sum=0;
00346     int border=0;
00347 
00348     tmp=out=*out_param;
00349     in =  *in_param;
00350 
00351     do{
00352         int ret, size, consumed;
00353         if(!s->resample_in_constraint && s->in_buffer_count){
00354             buf_set(&tmp, &s->in_buffer, s->in_buffer_index);
00355             ret= swri_multiple_resample(s->resample, &out, out_count, &tmp, s->in_buffer_count, &consumed);
00356             out_count -= ret;
00357             ret_sum += ret;
00358             buf_set(&out, &out, ret);
00359             s->in_buffer_count -= consumed;
00360             s->in_buffer_index += consumed;
00361 
00362             if(!in_count)
00363                 break;
00364             if(s->in_buffer_count <= border){
00365                 buf_set(&in, &in, -s->in_buffer_count);
00366                 in_count += s->in_buffer_count;
00367                 s->in_buffer_count=0;
00368                 s->in_buffer_index=0;
00369                 border = 0;
00370             }
00371         }
00372 
00373         if(in_count && !s->in_buffer_count){
00374             s->in_buffer_index=0;
00375             ret= swri_multiple_resample(s->resample, &out, out_count, &in, in_count, &consumed);
00376             out_count -= ret;
00377             ret_sum += ret;
00378             buf_set(&out, &out, ret);
00379             in_count -= consumed;
00380             buf_set(&in, &in, consumed);
00381         }
00382 
00383         //TODO is this check sane considering the advanced copy avoidance below
00384         size= s->in_buffer_index + s->in_buffer_count + in_count;
00385         if(   size > s->in_buffer.count
00386            && s->in_buffer_count + in_count <= s->in_buffer_index){
00387             buf_set(&tmp, &s->in_buffer, s->in_buffer_index);
00388             copy(&s->in_buffer, &tmp, s->in_buffer_count);
00389             s->in_buffer_index=0;
00390         }else
00391             if((ret=realloc_audio(&s->in_buffer, size)) < 0)
00392                 return ret;
00393 
00394         if(in_count){
00395             int count= in_count;
00396             if(s->in_buffer_count && s->in_buffer_count+2 < count && out_count) count= s->in_buffer_count+2;
00397 
00398             buf_set(&tmp, &s->in_buffer, s->in_buffer_index + s->in_buffer_count);
00399             copy(&tmp, &in, /*in_*/count);
00400             s->in_buffer_count += count;
00401             in_count -= count;
00402             border += count;
00403             buf_set(&in, &in, count);
00404             s->resample_in_constraint= 0;
00405             if(s->in_buffer_count != count || in_count)
00406                 continue;
00407         }
00408         break;
00409     }while(1);
00410 
00411     s->resample_in_constraint= !!out_count;
00412 
00413     return ret_sum;
00414 }
00415 
00416 int swr_convert(struct SwrContext *s, uint8_t *out_arg[SWR_CH_MAX], int out_count,
00417                          const uint8_t *in_arg [SWR_CH_MAX], int  in_count){
00418     AudioData *postin, *midbuf, *preout;
00419     int ret/*, in_max*/;
00420     AudioData * in= &s->in;
00421     AudioData *out= &s->out;
00422     AudioData preout_tmp, midbuf_tmp;
00423 
00424     if(!s->resample){
00425         if(in_count > out_count)
00426             return -1;
00427         out_count = in_count;
00428     }
00429 
00430     if(!in_arg){
00431         if(s->in_buffer_count){
00432             AudioData *a= &s->in_buffer;
00433             int i, j, ret;
00434             if((ret=realloc_audio(a, s->in_buffer_index + 2*s->in_buffer_count)) < 0)
00435                 return ret;
00436             av_assert0(a->planar);
00437             for(i=0; i<a->ch_count; i++){
00438                 for(j=0; j<s->in_buffer_count; j++){
00439                     memcpy(a->ch[i] + (s->in_buffer_index+s->in_buffer_count+j  )*a->bps,
00440                            a->ch[i] + (s->in_buffer_index+s->in_buffer_count-j-1)*a->bps, a->bps);
00441                 }
00442             }
00443             s->in_buffer_count += (s->in_buffer_count+1)/2;
00444             s->resample_in_constraint = 0;
00445         }else{
00446             return 0;
00447         }
00448     }else
00449         fill_audiodata(in ,  (void*)in_arg);
00450     fill_audiodata(out, out_arg);
00451 
00452     if(s->full_convert){
00453         av_assert0(!s->resample);
00454         swri_audio_convert(s->full_convert, out, in, in_count);
00455         return out_count;
00456     }
00457 
00458 //     in_max= out_count*(int64_t)s->in_sample_rate / s->out_sample_rate + resample_filter_taps;
00459 //     in_count= FFMIN(in_count, in_in + 2 - s->hist_buffer_count);
00460 
00461     if((ret=realloc_audio(&s->postin, in_count))<0)
00462         return ret;
00463     if(s->resample_first){
00464         av_assert0(s->midbuf.ch_count == s->used_ch_count);
00465         if((ret=realloc_audio(&s->midbuf, out_count))<0)
00466             return ret;
00467     }else{
00468         av_assert0(s->midbuf.ch_count ==  s->out.ch_count);
00469         if((ret=realloc_audio(&s->midbuf,  in_count))<0)
00470             return ret;
00471     }
00472     if((ret=realloc_audio(&s->preout, out_count))<0)
00473         return ret;
00474 
00475     postin= &s->postin;
00476 
00477     midbuf_tmp= s->midbuf;
00478     midbuf= &midbuf_tmp;
00479     preout_tmp= s->preout;
00480     preout= &preout_tmp;
00481 
00482     if(s->int_sample_fmt == s-> in_sample_fmt && s->in.planar)
00483         postin= in;
00484 
00485     if(s->resample_first ? !s->resample : !s->rematrix)
00486         midbuf= postin;
00487 
00488     if(s->resample_first ? !s->rematrix : !s->resample)
00489         preout= midbuf;
00490 
00491     if(s->int_sample_fmt == s->out_sample_fmt && s->out.planar){
00492         if(preout==in){
00493             out_count= FFMIN(out_count, in_count); //TODO check at teh end if this is needed or redundant
00494             av_assert0(s->in.planar); //we only support planar internally so it has to be, we support copying non planar though
00495             copy(out, in, out_count);
00496             return out_count;
00497         }
00498         else if(preout==postin) preout= midbuf= postin= out;
00499         else if(preout==midbuf) preout= midbuf= out;
00500         else                    preout= out;
00501     }
00502 
00503     if(in != postin){
00504         swri_audio_convert(s->in_convert, postin, in, in_count);
00505     }
00506 
00507     if(s->resample_first){
00508         if(postin != midbuf)
00509             out_count= resample(s, midbuf, out_count, postin, in_count);
00510         if(midbuf != preout)
00511             swri_rematrix(s, preout, midbuf, out_count, preout==out);
00512     }else{
00513         if(postin != midbuf)
00514             swri_rematrix(s, midbuf, postin, in_count, midbuf==out);
00515         if(midbuf != preout)
00516             out_count= resample(s, preout, out_count, midbuf, in_count);
00517     }
00518 
00519     if(preout != out){
00520 //FIXME packed doesnt need more than 1 chan here!
00521         swri_audio_convert(s->out_convert, out, preout, out_count);
00522     }
00523     if(!in_arg)
00524         s->in_buffer_count = 0;
00525     return out_count;
00526 }
00527