libx264: list possible profiles.
[ffmpeg.git] / libswresample / swresample.c
index b0a0041..89fe0f7 100644 (file)
@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2011 Michael Niedermayer (michaelni@gmx.at)
+ * Copyright (C) 2011-2012 Michael Niedermayer (michaelni@gmx.at)
  *
  * This file is part of libswresample
  *
@@ -24,6 +24,8 @@
 #include "libavutil/avassert.h"
 #include "libavutil/audioconvert.h"
 
+#include <float.h>
+
 #define  C30DB  M_SQRT2
 #define  C15DB  1.189207115
 #define C__0DB  1.0
 #define C_45DB  0.594603558
 #define C_60DB  0.5
 
+#define ALIGN 32
 
 //TODO split options array out?
 #define OFFSET(x) offsetof(SwrContext,x)
+#define PARAM AV_OPT_FLAG_AUDIO_PARAM
+
 static const AVOption options[]={
-{"ich",  "input channel count", OFFSET( in.ch_count   ), AV_OPT_TYPE_INT, {.dbl=2}, 1, SWR_CH_MAX, 0},
-{"och", "output channel count", OFFSET(out.ch_count   ), AV_OPT_TYPE_INT, {.dbl=2}, 1, SWR_CH_MAX, 0},
-{"uch",   "used channel count", OFFSET(used_ch_count  ), AV_OPT_TYPE_INT, {.dbl=0}, 0, SWR_CH_MAX, 0},
-{"isr",  "input sample rate"  , OFFSET( in_sample_rate), AV_OPT_TYPE_INT, {.dbl=48000}, 1, INT_MAX, 0},
-{"osr", "output sample rate"  , OFFSET(out_sample_rate), AV_OPT_TYPE_INT, {.dbl=48000}, 1, INT_MAX, 0},
-//{"ip" ,  "input planar"       , OFFSET( in.planar     ), AV_OPT_TYPE_INT, {.dbl=0},    0,       1, 0},
-//{"op" , "output planar"       , OFFSET(out.planar     ), AV_OPT_TYPE_INT, {.dbl=0},    0,       1, 0},
-{"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},
-{"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},
-{"tsf", "internal sample format", OFFSET(int_sample_fmt ), AV_OPT_TYPE_INT, {.dbl=AV_SAMPLE_FMT_NONE}, -1, AV_SAMPLE_FMT_FLT, 0},
-{"icl",  "input channel layout" , OFFSET( in_ch_layout), AV_OPT_TYPE_INT64, {.dbl=0}, 0, INT64_MAX, 0, "channel_layout"},
-{"ocl",  "output channel layout", OFFSET(out_ch_layout), AV_OPT_TYPE_INT64, {.dbl=0}, 0, INT64_MAX, 0, "channel_layout"},
-{"clev", "center mix level"     , OFFSET(clev)         , AV_OPT_TYPE_FLOAT, {.dbl=C_30DB}, 0, 4, 0},
-{"slev", "sourround mix level"  , OFFSET(slev)         , AV_OPT_TYPE_FLOAT, {.dbl=C_30DB}, 0, 4, 0},
-{"rmvol", "rematrix volume"     , OFFSET(rematrix_volume), AV_OPT_TYPE_FLOAT, {.dbl=1.0}, -1000, 1000, 0},
-{"flags", NULL                  , OFFSET(flags)        , AV_OPT_TYPE_FLAGS, {.dbl=0}, 0,  UINT_MAX, 0, "flags"},
-{"res", "force resampling", 0, AV_OPT_TYPE_CONST, {.dbl=SWR_FLAG_RESAMPLE}, INT_MIN, INT_MAX, 0, "flags"},
+{"ich"                  ,  "Input Channel Count"        , OFFSET( in.ch_count   ), AV_OPT_TYPE_INT  , {.dbl=2                     }, 0      , SWR_CH_MAX, PARAM},
+{"in_channel_count"     ,  "Input Channel Count"        , OFFSET( in.ch_count   ), AV_OPT_TYPE_INT  , {.dbl=2                     }, 0      , SWR_CH_MAX, PARAM},
+{"och"                  , "Output Channel Count"        , OFFSET(out.ch_count   ), AV_OPT_TYPE_INT  , {.dbl=2                     }, 0      , SWR_CH_MAX, PARAM},
+{"out_channel_count"    , "Output Channel Count"        , OFFSET(out.ch_count   ), AV_OPT_TYPE_INT  , {.dbl=2                     }, 0      , SWR_CH_MAX, PARAM},
+{"uch"                  ,   "Used Channel Count"        , OFFSET(used_ch_count  ), AV_OPT_TYPE_INT  , {.dbl=0                     }, 0      , SWR_CH_MAX, PARAM},
+{"used_channel_count"   ,   "Used Channel Count"        , OFFSET(used_ch_count  ), AV_OPT_TYPE_INT  , {.dbl=0                     }, 0      , SWR_CH_MAX, PARAM},
+{"isr"                  ,  "Input Sample Rate"          , OFFSET( in_sample_rate), AV_OPT_TYPE_INT  , {.dbl=0                     }, 0      , INT_MAX   , PARAM},
+{"in_sample_rate"       ,  "Input Sample Rate"          , OFFSET( in_sample_rate), AV_OPT_TYPE_INT  , {.dbl=0                     }, 0      , INT_MAX   , PARAM},
+{"osr"                  , "Output Sample Rate"          , OFFSET(out_sample_rate), AV_OPT_TYPE_INT  , {.dbl=0                     }, 0      , INT_MAX   , PARAM},
+{"out_sample_rate"      , "Output Sample Rate"          , OFFSET(out_sample_rate), AV_OPT_TYPE_INT  , {.dbl=0                     }, 0      , INT_MAX   , PARAM},
+{"isf"                  ,    "Input Sample Format"      , OFFSET( in_sample_fmt ), AV_OPT_TYPE_INT  , {.dbl=AV_SAMPLE_FMT_NONE    }, -1     , AV_SAMPLE_FMT_NB-1+256, PARAM},
+{"in_sample_fmt"        ,    "Input Sample Format"      , OFFSET( in_sample_fmt ), AV_OPT_TYPE_INT  , {.dbl=AV_SAMPLE_FMT_NONE    }, -1     , AV_SAMPLE_FMT_NB-1+256, PARAM},
+{"osf"                  ,   "Output Sample Format"      , OFFSET(out_sample_fmt ), AV_OPT_TYPE_INT  , {.dbl=AV_SAMPLE_FMT_NONE    }, -1     , AV_SAMPLE_FMT_NB-1+256, PARAM},
+{"out_sample_fmt"       ,   "Output Sample Format"      , OFFSET(out_sample_fmt ), AV_OPT_TYPE_INT  , {.dbl=AV_SAMPLE_FMT_NONE    }, -1     , AV_SAMPLE_FMT_NB-1+256, PARAM},
+{"tsf"                  , "Internal Sample Format"      , OFFSET(int_sample_fmt ), AV_OPT_TYPE_INT  , {.dbl=AV_SAMPLE_FMT_NONE    }, -1     , AV_SAMPLE_FMT_FLTP, PARAM},
+{"internal_sample_fmt"  , "Internal Sample Format"      , OFFSET(int_sample_fmt ), AV_OPT_TYPE_INT  , {.dbl=AV_SAMPLE_FMT_NONE    }, -1     , AV_SAMPLE_FMT_FLTP, PARAM},
+{"icl"                  ,   "Input Channel Layout"      , OFFSET( in_ch_layout  ), AV_OPT_TYPE_INT64, {.dbl=0                     }, 0      , INT64_MAX , PARAM, "channel_layout"},
+{"in_channel_layout"    ,   "Input Channel Layout"      , OFFSET( in_ch_layout  ), AV_OPT_TYPE_INT64, {.dbl=0                     }, 0      , INT64_MAX , PARAM, "channel_layout"},
+{"ocl"                  ,  "Output Channel Layout"      , OFFSET(out_ch_layout  ), AV_OPT_TYPE_INT64, {.dbl=0                     }, 0      , INT64_MAX , PARAM, "channel_layout"},
+{"out_channel_layout"   ,  "Output Channel Layout"      , OFFSET(out_ch_layout  ), AV_OPT_TYPE_INT64, {.dbl=0                     }, 0      , INT64_MAX , PARAM, "channel_layout"},
+{"clev"                 ,    "Center Mix Level"         , OFFSET(clev           ), AV_OPT_TYPE_FLOAT, {.dbl=C_30DB                }, -32    , 32        , PARAM},
+{"center_mix_level"     ,    "Center Mix Level"         , OFFSET(clev           ), AV_OPT_TYPE_FLOAT, {.dbl=C_30DB                }, -32    , 32        , PARAM},
+{"slev"                 , "Sourround Mix Level"         , OFFSET(slev           ), AV_OPT_TYPE_FLOAT, {.dbl=C_30DB                }, -32    , 32        , PARAM},
+{"surround_mix_level"   , "Sourround Mix Level"         , OFFSET(slev           ), AV_OPT_TYPE_FLOAT, {.dbl=C_30DB                }, -32    , 32        , PARAM},
+{"lfe_mix_level"        , "LFE Mix Level"               , OFFSET(lfe_mix_level  ), AV_OPT_TYPE_FLOAT, {.dbl=0                     }, -32    , 32        , PARAM},
+{"rmvol"                , "Rematrix Volume"             , OFFSET(rematrix_volume), AV_OPT_TYPE_FLOAT, {.dbl=1.0                   }, -1000  , 1000      , PARAM},
+{"rematrix_volume"      , "Rematrix Volume"             , OFFSET(rematrix_volume), AV_OPT_TYPE_FLOAT, {.dbl=1.0                   }, -1000  , 1000      , PARAM},
+{"flags"                , NULL                          , OFFSET(flags          ), AV_OPT_TYPE_FLAGS, {.dbl=0                     }, 0      , UINT_MAX  , PARAM, "flags"},
+{"swr_flags"            , NULL                          , OFFSET(flags          ), AV_OPT_TYPE_FLAGS, {.dbl=0                     }, 0      , UINT_MAX  , PARAM, "flags"},
+{"res"                  , "Force Resampling"            , 0                      , AV_OPT_TYPE_CONST, {.dbl=SWR_FLAG_RESAMPLE     }, INT_MIN, INT_MAX   , PARAM, "flags"},
+{"dither_scale"         , "Dither Scale"                , OFFSET(dither_scale   ), AV_OPT_TYPE_FLOAT, {.dbl=1                     }, 0      , INT_MAX   , PARAM},
+{"dither_method"        , "Dither Method"               , OFFSET(dither_method  ), AV_OPT_TYPE_INT  , {.dbl=0                     }, 0      , SWR_DITHER_NB-1, PARAM, "dither_method"},
+{"rectangular"          , "Rectangular Dither"          , 0                      , AV_OPT_TYPE_CONST, {.dbl=SWR_DITHER_RECTANGULAR}, INT_MIN, INT_MAX   , PARAM, "dither_method"},
+{"triangular"           ,  "Triangular Dither"          , 0                      , AV_OPT_TYPE_CONST, {.dbl=SWR_DITHER_TRIANGULAR }, INT_MIN, INT_MAX   , PARAM, "dither_method"},
+{"triangular_hp"        , "Triangular Dither With High Pass" , 0                 , AV_OPT_TYPE_CONST, {.dbl=SWR_DITHER_TRIANGULAR_HIGHPASS }, INT_MIN, INT_MAX, PARAM, "dither_method"},
+{"filter_size"          , "Resampling Filter Size"      , OFFSET(filter_size)    , AV_OPT_TYPE_INT  , {.dbl=16                    }, 0      , INT_MAX   , PARAM },
+{"phase_shift"          , "Resampling Phase Shift"      , OFFSET(phase_shift)    , AV_OPT_TYPE_INT  , {.dbl=10                    }, 0      , 30        , PARAM },
+{"linear_interp"        , "Use Linear Interpolation"    , OFFSET(linear_interp)  , AV_OPT_TYPE_INT  , {.dbl=0                     }, 0      , 1         , PARAM },
+{"cutoff"               , "Cutoff Frequency Ratio"      , OFFSET(cutoff)         , AV_OPT_TYPE_DOUBLE,{.dbl=0.8                   }, 0      , 1         , PARAM },
+{"min_comp"             , "Minimum difference between timestamps and audio data (in seconds) below which no timestamp compensation of either kind is applied"
+                                                        , OFFSET(min_compensation),AV_OPT_TYPE_FLOAT ,{.dbl=FLT_MAX               }, 0      , FLT_MAX   , PARAM },
+{"min_hard_comp"        , "Minimum difference between timestamps and audio data (in seconds) to trigger padding/trimming the data."
+                                                   , OFFSET(min_hard_compensation),AV_OPT_TYPE_FLOAT ,{.dbl=0.1                   }, 0      , INT_MAX   , PARAM },
+{"comp_duration"        , "Duration (in seconds) over which data is stretched/squeezed to make it match the timestamps."
+                                              , OFFSET(soft_compensation_duration),AV_OPT_TYPE_FLOAT ,{.dbl=1                     }, 0      , INT_MAX   , PARAM },
+{"max_soft_comp"        , "Maximum factor by which data is stretched/squeezed to make it match the timestamps."
+                                                   , OFFSET(max_soft_compensation),AV_OPT_TYPE_FLOAT ,{.dbl=0                     }, INT_MIN, INT_MAX   , PARAM },
+{ "filter_type"         , "Filter Type"                 , OFFSET(filter_type)    , AV_OPT_TYPE_INT  , { SWR_FILTER_TYPE_KAISER }, SWR_FILTER_TYPE_CUBIC, SWR_FILTER_TYPE_KAISER, PARAM, "filter_type" },
+    { "cubic"           , "Cubic"                       , 0                      , AV_OPT_TYPE_CONST, { SWR_FILTER_TYPE_CUBIC            }, INT_MIN, INT_MAX, PARAM, "filter_type" },
+    { "blackman_nuttall", "Blackman Nuttall Windowed Sinc", 0                    , AV_OPT_TYPE_CONST, { SWR_FILTER_TYPE_BLACKMAN_NUTTALL }, INT_MIN, INT_MAX, PARAM, "filter_type" },
+    { "kaiser"          , "Kaiser Windowed Sinc"        , 0                      , AV_OPT_TYPE_CONST, { SWR_FILTER_TYPE_KAISER           }, INT_MIN, INT_MAX, PARAM, "filter_type" },
+{ "kaiser_beta"         , "Kaiser Window Beta"          ,OFFSET(kaiser_beta)     , AV_OPT_TYPE_INT  , {.dbl=9                     }, 2      , 16        , PARAM },
 
 {0}
 };
@@ -62,16 +102,31 @@ static const char* context_to_name(void* ptr) {
 }
 
 static const AVClass av_class = {
-    .class_name                = "SwrContext",
+    .class_name                = "SWResampler",
     .item_name                 = context_to_name,
     .option                    = options,
     .version                   = LIBAVUTIL_VERSION_INT,
     .log_level_offset_offset   = OFFSET(log_level_offset),
     .parent_log_context_offset = OFFSET(log_ctx),
+    .category                  = AV_CLASS_CATEGORY_SWRESAMPLER,
 };
 
-static int resample(SwrContext *s, AudioData *out_param, int out_count,
-                             const AudioData * in_param, int in_count);
+unsigned swresample_version(void)
+{
+    av_assert0(LIBSWRESAMPLE_VERSION_MICRO >= 100);
+    return LIBSWRESAMPLE_VERSION_INT;
+}
+
+const char *swresample_configuration(void)
+{
+    return FFMPEG_CONFIGURATION;
+}
+
+const char *swresample_license(void)
+{
+#define LICENSE_PREFIX "libswresample license: "
+    return LICENSE_PREFIX FFMPEG_LICENSE + sizeof(LICENSE_PREFIX) - 1;
+}
 
 int swr_set_channel_mapping(struct SwrContext *s, const int *channel_map){
     if(!s || s->in_convert) // s needs to be allocated but not initialized
@@ -80,6 +135,11 @@ int swr_set_channel_mapping(struct SwrContext *s, const int *channel_map){
     return 0;
 }
 
+const AVClass *swr_get_class(void)
+{
+    return &av_class;
+}
+
 struct SwrContext *swr_alloc(void){
     SwrContext *s= av_mallocz(sizeof(SwrContext));
     if(s){
@@ -105,12 +165,18 @@ struct SwrContext *swr_alloc_set_opts(struct SwrContext *s,
     av_opt_set_int(s, "icl", in_ch_layout,    0);
     av_opt_set_int(s, "isf", in_sample_fmt,   0);
     av_opt_set_int(s, "isr", in_sample_rate,  0);
-    av_opt_set_int(s, "tsf", AV_SAMPLE_FMT_S16, 0);
+    av_opt_set_int(s, "tsf", AV_SAMPLE_FMT_NONE,   0);
     av_opt_set_int(s, "ich", av_get_channel_layout_nb_channels(s-> in_ch_layout), 0);
     av_opt_set_int(s, "och", av_get_channel_layout_nb_channels(s->out_ch_layout), 0);
+    av_opt_set_int(s, "uch", 0, 0);
     return s;
 }
 
+static void set_audiodata_fmt(AudioData *a, enum AVSampleFormat fmt){
+    a->fmt   = fmt;
+    a->bps   = av_get_bytes_per_sample(fmt);
+    a->planar= av_sample_fmt_is_planar(fmt);
+}
 
 static void free_temp(AudioData *a){
     av_free(a->data);
@@ -124,10 +190,12 @@ void swr_free(SwrContext **ss){
         free_temp(&s->midbuf);
         free_temp(&s->preout);
         free_temp(&s->in_buffer);
+        free_temp(&s->dither);
         swri_audio_convert_free(&s-> in_convert);
         swri_audio_convert_free(&s->out_convert);
         swri_audio_convert_free(&s->full_convert);
         swri_resample_free(&s->resample);
+        swri_rematrix_free(s);
     }
 
     av_freep(ss);
@@ -141,43 +209,55 @@ int swr_init(struct SwrContext *s){
     free_temp(&s->midbuf);
     free_temp(&s->preout);
     free_temp(&s->in_buffer);
+    free_temp(&s->dither);
     swri_audio_convert_free(&s-> in_convert);
     swri_audio_convert_free(&s->out_convert);
     swri_audio_convert_free(&s->full_convert);
+    swri_rematrix_free(s);
 
-    s-> in.planar= s-> in_sample_fmt >= 0x100;
-    s->out.planar= s->out_sample_fmt >= 0x100;
-    s-> in_sample_fmt &= 0xFF;
-    s->out_sample_fmt &= 0xFF;
+    s->flushed = 0;
 
     if(s-> in_sample_fmt >= AV_SAMPLE_FMT_NB){
-        av_log(s, AV_LOG_ERROR, "Requested sample format %s is invalid\n", av_get_sample_fmt_name(s->in_sample_fmt));
+        av_log(s, AV_LOG_ERROR, "Requested input sample format %d is invalid\n", s->in_sample_fmt);
         return AVERROR(EINVAL);
     }
     if(s->out_sample_fmt >= AV_SAMPLE_FMT_NB){
-        av_log(s, AV_LOG_ERROR, "Requested sample format %s is invalid\n", av_get_sample_fmt_name(s->out_sample_fmt));
+        av_log(s, AV_LOG_ERROR, "Requested output sample format %d is invalid\n", s->out_sample_fmt);
         return AVERROR(EINVAL);
     }
 
-    if(   s->int_sample_fmt != AV_SAMPLE_FMT_S16
-        &&s->int_sample_fmt != AV_SAMPLE_FMT_FLT){
-        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));
-        return AVERROR(EINVAL);
+    if(s->int_sample_fmt == AV_SAMPLE_FMT_NONE){
+        if(av_get_planar_sample_fmt(s->in_sample_fmt) <= AV_SAMPLE_FMT_S16P){
+            s->int_sample_fmt= AV_SAMPLE_FMT_S16P;
+        }else if(av_get_planar_sample_fmt(s->in_sample_fmt) <= AV_SAMPLE_FMT_FLTP){
+            s->int_sample_fmt= AV_SAMPLE_FMT_FLTP;
+        }else{
+            av_log(s, AV_LOG_DEBUG, "Using double precision mode\n");
+            s->int_sample_fmt= AV_SAMPLE_FMT_DBLP;
+        }
     }
 
-    //FIXME should we allow/support using FLT on material that doesnt need it ?
-    if(s->in_sample_fmt <= AV_SAMPLE_FMT_S16 || s->int_sample_fmt==AV_SAMPLE_FMT_S16){
-        s->int_sample_fmt= AV_SAMPLE_FMT_S16;
-    }else
-        s->int_sample_fmt= AV_SAMPLE_FMT_FLT;
+    if(   s->int_sample_fmt != AV_SAMPLE_FMT_S16P
+        &&s->int_sample_fmt != AV_SAMPLE_FMT_S32P
+        &&s->int_sample_fmt != AV_SAMPLE_FMT_FLTP
+        &&s->int_sample_fmt != AV_SAMPLE_FMT_DBLP){
+        av_log(s, AV_LOG_ERROR, "Requested sample format %s is not supported internally, S16/S32/FLT/DBL is supported\n", av_get_sample_fmt_name(s->int_sample_fmt));
+        return AVERROR(EINVAL);
+    }
 
+    set_audiodata_fmt(&s-> in, s-> in_sample_fmt);
+    set_audiodata_fmt(&s->out, s->out_sample_fmt);
 
     if (s->out_sample_rate!=s->in_sample_rate || (s->flags & SWR_FLAG_RESAMPLE)){
-        s->resample = swri_resample_init(s->resample, s->out_sample_rate, s->in_sample_rate, 16, 10, 0, 0.8);
+        s->resample = swri_resample_init(s->resample, s->out_sample_rate, s->in_sample_rate, s->filter_size, s->phase_shift, s->linear_interp, s->cutoff, s->int_sample_fmt, s->filter_type, s->kaiser_beta);
     }else
         swri_resample_free(&s->resample);
-    if(s->int_sample_fmt != AV_SAMPLE_FMT_S16 && s->resample){
-        av_log(s, AV_LOG_ERROR, "Resampling only supported with internal s16 currently\n"); //FIXME
+    if(    s->int_sample_fmt != AV_SAMPLE_FMT_S16P
+        && s->int_sample_fmt != AV_SAMPLE_FMT_S32P
+        && s->int_sample_fmt != AV_SAMPLE_FMT_FLTP
+        && s->int_sample_fmt != AV_SAMPLE_FMT_DBLP
+        && s->resample){
+        av_log(s, AV_LOG_ERROR, "Resampling only supported with internal s16/s32/flt/dbl\n");
         return -1;
     }
 
@@ -194,7 +274,8 @@ int swr_init(struct SwrContext *s){
     if(!s->out_ch_layout)
         s->out_ch_layout= av_get_default_channel_layout(s->out.ch_count);
 
-    s->rematrix= s->out_ch_layout  !=s->in_ch_layout || s->rematrix_volume!=1.0;
+    s->rematrix= s->out_ch_layout  !=s->in_ch_layout || s->rematrix_volume!=1.0 ||
+                 s->rematrix_custom;
 
 #define RSC 1 //FIXME finetune
     if(!s-> in.ch_count)
@@ -204,16 +285,24 @@ int swr_init(struct SwrContext *s){
     if(!s->out.ch_count)
         s->out.ch_count= av_get_channel_layout_nb_channels(s->out_ch_layout);
 
-av_assert0(s-> in.ch_count);
+    if(!s-> in.ch_count){
+        av_assert0(!s->in_ch_layout);
+        av_log(s, AV_LOG_ERROR, "Input channel count and layout are unset\n");
+        return -1;
+    }
+
+    if ((!s->out_ch_layout || !s->in_ch_layout) && s->used_ch_count != s->out.ch_count && !s->rematrix_custom) {
+        av_log(s, AV_LOG_ERROR, "Rematrix is needed but there is not enough information to do it\n");
+        return -1;
+    }
+
 av_assert0(s->used_ch_count);
 av_assert0(s->out.ch_count);
     s->resample_first= RSC*s->out.ch_count/s->in.ch_count - RSC < s->out_sample_rate/(float)s-> in_sample_rate - 1.0;
 
-    s-> in.bps= av_get_bytes_per_sample(s-> in_sample_fmt);
-    s->int_bps= av_get_bytes_per_sample(s->int_sample_fmt);
-    s->out.bps= av_get_bytes_per_sample(s->out_sample_fmt);
+    s->in_buffer= s->in;
 
-    if(!s->resample && !s->rematrix && !s->channel_map){
+    if(!s->resample && !s->rematrix && !s->channel_map && !s->dither_method){
         s->full_convert = swri_audio_convert_alloc(s->out_sample_fmt,
                                                    s-> in_sample_fmt, s-> in.ch_count, NULL, 0);
         return 0;
@@ -228,22 +317,30 @@ av_assert0(s->out.ch_count);
     s->postin= s->in;
     s->preout= s->out;
     s->midbuf= s->in;
-    s->in_buffer= s->in;
+
     if(s->channel_map){
         s->postin.ch_count=
-        s->midbuf.ch_count=
-        s->in_buffer.ch_count= s->used_ch_count;
+        s->midbuf.ch_count= s->used_ch_count;
+        if(s->resample)
+            s->in_buffer.ch_count= s->used_ch_count;
     }
     if(!s->resample_first){
         s->midbuf.ch_count= s->out.ch_count;
-        s->in_buffer.ch_count = s->out.ch_count;
+        if(s->resample)
+            s->in_buffer.ch_count = s->out.ch_count;
     }
 
-    s->in_buffer.bps = s->postin.bps = s->midbuf.bps = s->preout.bps =  s->int_bps;
-    s->in_buffer.planar = s->postin.planar = s->midbuf.planar = s->preout.planar =  1;
+    set_audiodata_fmt(&s->postin, s->int_sample_fmt);
+    set_audiodata_fmt(&s->midbuf, s->int_sample_fmt);
+    set_audiodata_fmt(&s->preout, s->int_sample_fmt);
+
+    if(s->resample){
+        set_audiodata_fmt(&s->in_buffer, s->int_sample_fmt);
+    }
 
+    s->dither = s->preout;
 
-    if(s->rematrix)
+    if(s->rematrix || s->dither_method)
         return swri_rematrix_init(s);
 
     return 0;
@@ -253,25 +350,28 @@ static int realloc_audio(AudioData *a, int count){
     int i, countb;
     AudioData old;
 
+    if(count < 0 || count > INT_MAX/2/a->bps/a->ch_count)
+        return AVERROR(EINVAL);
+
     if(a->count >= count)
         return 0;
 
     count*=2;
 
-    countb= FFALIGN(count*a->bps, 32);
+    countb= FFALIGN(count*a->bps, ALIGN);
     old= *a;
 
-    av_assert0(a->planar);
     av_assert0(a->bps);
     av_assert0(a->ch_count);
 
-    a->data= av_malloc(countb*a->ch_count);
+    a->data= av_mallocz(countb*a->ch_count);
     if(!a->data)
         return AVERROR(ENOMEM);
     for(i=0; i<a->ch_count; i++){
         a->ch[i]= a->data + i*(a->planar ? countb : a->bps);
         if(a->planar) memcpy(a->ch[i], old.ch[i], a->count*a->bps);
     }
+    if(!a->planar) memcpy(a->ch[0], old.ch[0], a->count*a->ch_count*a->bps);
     av_free(old.data);
     a->count= count;
 
@@ -293,7 +393,9 @@ static void copy(AudioData *out, AudioData *in,
 
 static void fill_audiodata(AudioData *out, uint8_t *in_arg [SWR_CH_MAX]){
     int i;
-    if(out->planar){
+    if(!in_arg){
+        memset(out->ch, 0, sizeof(out->ch));
+    }else if(out->planar){
         for(i=0; i<out->ch_count; i++)
             out->ch[i]= in_arg[i];
     }else{
@@ -302,41 +404,118 @@ static void fill_audiodata(AudioData *out, uint8_t *in_arg [SWR_CH_MAX]){
     }
 }
 
-int swr_convert(struct SwrContext *s, uint8_t *out_arg[SWR_CH_MAX], int out_count,
-                         const uint8_t *in_arg [SWR_CH_MAX], int  in_count){
-    AudioData *postin, *midbuf, *preout;
-    int ret/*, in_max*/;
-    AudioData * in= &s->in;
-    AudioData *out= &s->out;
-    AudioData preout_tmp, midbuf_tmp;
+static void reversefill_audiodata(AudioData *out, uint8_t *in_arg [SWR_CH_MAX]){
+    int i;
+    if(out->planar){
+        for(i=0; i<out->ch_count; i++)
+            in_arg[i]= out->ch[i];
+    }else{
+        in_arg[0]= out->ch[0];
+    }
+}
 
-    if(!s->resample){
-        if(in_count > out_count)
-            return -1;
-        out_count = in_count;
+/**
+ *
+ * out may be equal in.
+ */
+static void buf_set(AudioData *out, AudioData *in, int count){
+    int ch;
+    if(in->planar){
+        for(ch=0; ch<out->ch_count; ch++)
+            out->ch[ch]= in->ch[ch] + count*out->bps;
+    }else{
+        for(ch=out->ch_count-1; ch>=0; ch--)
+            out->ch[ch]= in->ch[0] + (ch + count*out->ch_count) * out->bps;
     }
+}
 
-    if(!in_arg){
-        if(s->in_buffer_count){
-            AudioData *a= &s->in_buffer;
-            int i, j, ret;
-            if((ret=realloc_audio(a, s->in_buffer_index + 2*s->in_buffer_count)) < 0)
-                return ret;
-            av_assert0(a->planar);
-            for(i=0; i<a->ch_count; i++){
-                for(j=0; j<s->in_buffer_count; j++){
-                    memcpy(a->ch[i] + (s->in_buffer_index+s->in_buffer_count+j  )*a->bps,
-                           a->ch[i] + (s->in_buffer_index+s->in_buffer_count-j-1)*a->bps, a->bps);
-                }
+/**
+ *
+ * @return number of samples output per channel
+ */
+static int resample(SwrContext *s, AudioData *out_param, int out_count,
+                             const AudioData * in_param, int in_count){
+    AudioData in, out, tmp;
+    int ret_sum=0;
+    int border=0;
+
+    av_assert1(s->in_buffer.ch_count == in_param->ch_count);
+    av_assert1(s->in_buffer.planar   == in_param->planar);
+    av_assert1(s->in_buffer.fmt      == in_param->fmt);
+
+    tmp=out=*out_param;
+    in =  *in_param;
+
+    do{
+        int ret, size, consumed;
+        if(!s->resample_in_constraint && s->in_buffer_count){
+            buf_set(&tmp, &s->in_buffer, s->in_buffer_index);
+            ret= swri_multiple_resample(s->resample, &out, out_count, &tmp, s->in_buffer_count, &consumed);
+            out_count -= ret;
+            ret_sum += ret;
+            buf_set(&out, &out, ret);
+            s->in_buffer_count -= consumed;
+            s->in_buffer_index += consumed;
+
+            if(!in_count)
+                break;
+            if(s->in_buffer_count <= border){
+                buf_set(&in, &in, -s->in_buffer_count);
+                in_count += s->in_buffer_count;
+                s->in_buffer_count=0;
+                s->in_buffer_index=0;
+                border = 0;
             }
-            s->in_buffer_count += (s->in_buffer_count+1)/2;
-            s->resample_in_constraint = 0;
-        }else{
-            return 0;
         }
-    }else
-        fill_audiodata(in ,  (void*)in_arg);
-    fill_audiodata(out, out_arg);
+
+        if(in_count && !s->in_buffer_count){
+            s->in_buffer_index=0;
+            ret= swri_multiple_resample(s->resample, &out, out_count, &in, in_count, &consumed);
+            out_count -= ret;
+            ret_sum += ret;
+            buf_set(&out, &out, ret);
+            in_count -= consumed;
+            buf_set(&in, &in, consumed);
+        }
+
+        //TODO is this check sane considering the advanced copy avoidance below
+        size= s->in_buffer_index + s->in_buffer_count + in_count;
+        if(   size > s->in_buffer.count
+           && s->in_buffer_count + in_count <= s->in_buffer_index){
+            buf_set(&tmp, &s->in_buffer, s->in_buffer_index);
+            copy(&s->in_buffer, &tmp, s->in_buffer_count);
+            s->in_buffer_index=0;
+        }else
+            if((ret=realloc_audio(&s->in_buffer, size)) < 0)
+                return ret;
+
+        if(in_count){
+            int count= in_count;
+            if(s->in_buffer_count && s->in_buffer_count+2 < count && out_count) count= s->in_buffer_count+2;
+
+            buf_set(&tmp, &s->in_buffer, s->in_buffer_index + s->in_buffer_count);
+            copy(&tmp, &in, /*in_*/count);
+            s->in_buffer_count += count;
+            in_count -= count;
+            border += count;
+            buf_set(&in, &in, count);
+            s->resample_in_constraint= 0;
+            if(s->in_buffer_count != count || in_count)
+                continue;
+        }
+        break;
+    }while(1);
+
+    s->resample_in_constraint= !!out_count;
+
+    return ret_sum;
+}
+
+static int swr_convert_internal(struct SwrContext *s, AudioData *out, int out_count,
+                                                      AudioData *in , int  in_count){
+    AudioData *postin, *midbuf, *preout;
+    int ret/*, in_max*/;
+    AudioData preout_tmp, midbuf_tmp;
 
     if(s->full_convert){
         av_assert0(!s->resample);
@@ -368,7 +547,7 @@ int swr_convert(struct SwrContext *s, uint8_t *out_arg[SWR_CH_MAX], int out_coun
     preout_tmp= s->preout;
     preout= &preout_tmp;
 
-    if(s->int_sample_fmt == s-> in_sample_fmt && s->in.planar)
+    if(s->int_sample_fmt == s-> in_sample_fmt && s->in.planar && !s->channel_map)
         postin= in;
 
     if(s->resample_first ? !s->resample : !s->rematrix)
@@ -379,7 +558,7 @@ int swr_convert(struct SwrContext *s, uint8_t *out_arg[SWR_CH_MAX], int out_coun
 
     if(s->int_sample_fmt == s->out_sample_fmt && s->out.planar){
         if(preout==in){
-            out_count= FFMIN(out_count, in_count); //TODO check at teh end if this is needed or redundant
+            out_count= FFMIN(out_count, in_count); //TODO check at the end if this is needed or redundant
             av_assert0(s->in.planar); //we only support planar internally so it has to be, we support copying non planar though
             copy(out, in, out_count);
             return out_count;
@@ -405,102 +584,206 @@ int swr_convert(struct SwrContext *s, uint8_t *out_arg[SWR_CH_MAX], int out_coun
             out_count= resample(s, preout, out_count, midbuf, in_count);
     }
 
-    if(preout != out){
+    if(preout != out && out_count){
+        if(s->dither_method){
+            int ch;
+            int dither_count= FFMAX(out_count, 1<<16);
+            av_assert0(preout != in);
+
+            if((ret=realloc_audio(&s->dither, dither_count))<0)
+                return ret;
+            if(ret)
+                for(ch=0; ch<s->dither.ch_count; ch++)
+                    swri_get_dither(s, s->dither.ch[ch], s->dither.count, 12345678913579<<ch, s->out_sample_fmt, s->int_sample_fmt);
+            av_assert0(s->dither.ch_count == preout->ch_count);
+
+            if(s->dither_pos + out_count > s->dither.count)
+                s->dither_pos = 0;
+
+            for(ch=0; ch<preout->ch_count; ch++)
+                s->mix_2_1_f(preout->ch[ch], preout->ch[ch], s->dither.ch[ch] + s->dither.bps * s->dither_pos, s->native_one, 0, 0, out_count);
+
+            s->dither_pos += out_count;
+        }
 //FIXME packed doesnt need more than 1 chan here!
         swri_audio_convert(s->out_convert, out, preout, out_count);
     }
-    if(!in_arg)
-        s->in_buffer_count = 0;
     return out_count;
 }
 
-/**
- *
- * out may be equal in.
- */
-static void buf_set(AudioData *out, AudioData *in, int count){
-    if(in->planar){
-        int ch;
-        for(ch=0; ch<out->ch_count; ch++)
-            out->ch[ch]= in->ch[ch] + count*out->bps;
-    }else
-        out->ch[0]= in->ch[0] + count*out->ch_count*out->bps;
-}
+int swr_convert(struct SwrContext *s, uint8_t *out_arg[SWR_CH_MAX], int out_count,
+                                const uint8_t *in_arg [SWR_CH_MAX], int  in_count){
+    AudioData * in= &s->in;
+    AudioData *out= &s->out;
 
-/**
- *
- * @return number of samples output per channel
- */
-static int resample(SwrContext *s, AudioData *out_param, int out_count,
-                             const AudioData * in_param, int in_count){
-    AudioData in, out, tmp;
-    int ret_sum=0;
-    int border=0;
+    if(s->drop_output > 0){
+        int ret;
+        AudioData tmp = s->out;
+        uint8_t *tmp_arg[SWR_CH_MAX];
+        tmp.count = 0;
+        tmp.data  = NULL;
+        if((ret=realloc_audio(&tmp, s->drop_output))<0)
+            return ret;
 
-    tmp=out=*out_param;
-    in =  *in_param;
+        reversefill_audiodata(&tmp, tmp_arg);
+        s->drop_output *= -1; //FIXME find a less hackish solution
+        ret = swr_convert(s, tmp_arg, -s->drop_output, in_arg, in_count); //FIXME optimize but this is as good as never called so maybe it doesnt matter
+        s->drop_output *= -1;
+        if(ret>0)
+            s->drop_output -= ret;
 
-    do{
-        int ret, size, consumed;
-        if(!s->resample_in_constraint && s->in_buffer_count){
-            buf_set(&tmp, &s->in_buffer, s->in_buffer_index);
-            ret= swri_multiple_resample(s->resample, &out, out_count, &tmp, s->in_buffer_count, &consumed);
-            out_count -= ret;
-            ret_sum += ret;
-            buf_set(&out, &out, ret);
-            s->in_buffer_count -= consumed;
-            s->in_buffer_index += consumed;
+        av_freep(&tmp.data);
+        if(s->drop_output || !out_arg)
+            return 0;
+        in_count = 0;
+    }
 
-            if(!in_count)
-                break;
-            if(s->in_buffer_count <= border){
-                buf_set(&in, &in, -s->in_buffer_count);
-                in_count += s->in_buffer_count;
-                s->in_buffer_count=0;
-                s->in_buffer_index=0;
-                border = 0;
+    if(!in_arg){
+        if(s->in_buffer_count){
+            if (s->resample && !s->flushed) {
+                AudioData *a= &s->in_buffer;
+                int i, j, ret;
+                if((ret=realloc_audio(a, s->in_buffer_index + 2*s->in_buffer_count)) < 0)
+                    return ret;
+                av_assert0(a->planar);
+                for(i=0; i<a->ch_count; i++){
+                    for(j=0; j<s->in_buffer_count; j++){
+                        memcpy(a->ch[i] + (s->in_buffer_index+s->in_buffer_count+j  )*a->bps,
+                            a->ch[i] + (s->in_buffer_index+s->in_buffer_count-j-1)*a->bps, a->bps);
+                    }
+                }
+                s->in_buffer_count += (s->in_buffer_count+1)/2;
+                s->resample_in_constraint = 0;
+                s->flushed = 1;
             }
+        }else{
+            return 0;
         }
+    }else
+        fill_audiodata(in ,  (void*)in_arg);
 
-        if(in_count && !s->in_buffer_count){
-            s->in_buffer_index=0;
-            ret= swri_multiple_resample(s->resample, &out, out_count, &in, in_count, &consumed);
-            out_count -= ret;
-            ret_sum += ret;
-            buf_set(&out, &out, ret);
-            in_count -= consumed;
-            buf_set(&in, &in, consumed);
-        }
+    fill_audiodata(out, out_arg);
 
-        //TODO is this check sane considering the advanced copy avoidance below
-        size= s->in_buffer_index + s->in_buffer_count + in_count;
-        if(   size > s->in_buffer.count
-           && s->in_buffer_count + in_count <= s->in_buffer_index){
+    if(s->resample){
+        int ret = swr_convert_internal(s, out, out_count, in, in_count);
+        if(ret>0 && !s->drop_output)
+            s->outpts += ret * (int64_t)s->in_sample_rate;
+        return ret;
+    }else{
+        AudioData tmp= *in;
+        int ret2=0;
+        int ret, size;
+        size = FFMIN(out_count, s->in_buffer_count);
+        if(size){
             buf_set(&tmp, &s->in_buffer, s->in_buffer_index);
-            copy(&s->in_buffer, &tmp, s->in_buffer_count);
-            s->in_buffer_index=0;
-        }else
-            if((ret=realloc_audio(&s->in_buffer, size)) < 0)
+            ret= swr_convert_internal(s, out, size, &tmp, size);
+            if(ret<0)
                 return ret;
+            ret2= ret;
+            s->in_buffer_count -= ret;
+            s->in_buffer_index += ret;
+            buf_set(out, out, ret);
+            out_count -= ret;
+            if(!s->in_buffer_count)
+                s->in_buffer_index = 0;
+        }
 
         if(in_count){
-            int count= in_count;
-            if(s->in_buffer_count && s->in_buffer_count+2 < count && out_count) count= s->in_buffer_count+2;
+            size= s->in_buffer_index + s->in_buffer_count + in_count - out_count;
+
+            if(in_count > out_count) { //FIXME move after swr_convert_internal
+                if(   size > s->in_buffer.count
+                && s->in_buffer_count + in_count - out_count <= s->in_buffer_index){
+                    buf_set(&tmp, &s->in_buffer, s->in_buffer_index);
+                    copy(&s->in_buffer, &tmp, s->in_buffer_count);
+                    s->in_buffer_index=0;
+                }else
+                    if((ret=realloc_audio(&s->in_buffer, size)) < 0)
+                        return ret;
+            }
 
-            buf_set(&tmp, &s->in_buffer, s->in_buffer_index + s->in_buffer_count);
-            copy(&tmp, &in, /*in_*/count);
-            s->in_buffer_count += count;
-            in_count -= count;
-            border += count;
-            buf_set(&in, &in, count);
-            s->resample_in_constraint= 0;
-            if(s->in_buffer_count != count || in_count)
-                continue;
+            if(out_count){
+                size = FFMIN(in_count, out_count);
+                ret= swr_convert_internal(s, out, size, in, size);
+                if(ret<0)
+                    return ret;
+                buf_set(in, in, ret);
+                in_count -= ret;
+                ret2 += ret;
+            }
+            if(in_count){
+                buf_set(&tmp, &s->in_buffer, s->in_buffer_index + s->in_buffer_count);
+                copy(&tmp, in, in_count);
+                s->in_buffer_count += in_count;
+            }
         }
-        break;
-    }while(1);
+        if(ret2>0 && !s->drop_output)
+            s->outpts += ret2 * (int64_t)s->in_sample_rate;
+        return ret2;
+    }
+}
 
-    s->resample_in_constraint= !!out_count;
+int swr_drop_output(struct SwrContext *s, int count){
+    s->drop_output += count;
 
-    return ret_sum;
+    if(s->drop_output <= 0)
+        return 0;
+
+    av_log(s, AV_LOG_VERBOSE, "discarding %d audio samples\n", count);
+    return swr_convert(s, NULL, s->drop_output, NULL, 0);
+}
+
+int swr_inject_silence(struct SwrContext *s, int count){
+    int ret, i;
+    AudioData silence = s->out;
+    uint8_t *tmp_arg[SWR_CH_MAX];
+
+    if(count <= 0)
+        return 0;
+
+    silence.count = 0;
+    silence.data  = NULL;
+    if((ret=realloc_audio(&silence, count))<0)
+        return ret;
+
+    if(silence.planar) for(i=0; i<silence.ch_count; i++) {
+        memset(silence.ch[i], silence.bps==1 ? 0x80 : 0, count*silence.bps);
+    } else
+        memset(silence.ch[0], silence.bps==1 ? 0x80 : 0, count*silence.bps*silence.ch_count);
+
+    reversefill_audiodata(&silence, tmp_arg);
+    av_log(s, AV_LOG_VERBOSE, "adding %d audio samples of silence\n", count);
+    ret = swr_convert(s, NULL, 0, (const uint8_t**)tmp_arg, count);
+    av_freep(&silence.data);
+    return ret;
+}
+
+int64_t swr_next_pts(struct SwrContext *s, int64_t pts){
+    if(pts == INT64_MIN)
+        return s->outpts;
+    if(s->min_compensation >= FLT_MAX) {
+        return (s->outpts = pts - swr_get_delay(s, s->in_sample_rate * (int64_t)s->out_sample_rate));
+    } else {
+        int64_t delta = pts - swr_get_delay(s, s->in_sample_rate * (int64_t)s->out_sample_rate) - s->outpts;
+        double fdelta = delta /(double)(s->in_sample_rate * (int64_t)s->out_sample_rate);
+
+        if(fabs(fdelta) > s->min_compensation) {
+            if(!s->outpts || fabs(fdelta) > s->min_hard_compensation){
+                int ret;
+                if(delta > 0) ret = swr_inject_silence(s,  delta / s->out_sample_rate);
+                else          ret = swr_drop_output   (s, -delta / s-> in_sample_rate);
+                if(ret<0){
+                    av_log(s, AV_LOG_ERROR, "Failed to compensate for timestamp delta of %f\n", fdelta);
+                }
+            } else if(s->soft_compensation_duration && s->max_soft_compensation) {
+                int duration = s->out_sample_rate * s->soft_compensation_duration;
+                double max_soft_compensation = s->max_soft_compensation / (s->max_soft_compensation < 0 ? -s->in_sample_rate : 1);
+                int comp = av_clipf(fdelta, -max_soft_compensation, max_soft_compensation) * duration ;
+                av_log(s, AV_LOG_VERBOSE, "compensating audio timestamp drift:%f compensation:%d in:%d\n", fdelta, comp, duration);
+                swr_set_compensation(s, comp, duration);
+            }
+        }
+
+        return s->outpts;
+    }
 }