]> git.sesse.net Git - ffmpeg/blobdiff - ffmpeg.c
vf_fps: license cleanup
[ffmpeg] / ffmpeg.c
index 0bd318abc831e4c6cb058e3965ff0afe87809569..51cd58719826e9a157e69896fb65bc69d26bc8ed 100644 (file)
--- a/ffmpeg.c
+++ b/ffmpeg.c
@@ -35,6 +35,7 @@
 #include "libavformat/avformat.h"
 #include "libavdevice/avdevice.h"
 #include "libswscale/swscale.h"
+#include "libswresample/swresample.h"
 #include "libavutil/opt.h"
 #include "libavutil/audioconvert.h"
 #include "libavutil/parseutils.h"
 #include "libavutil/libm.h"
 #include "libavutil/imgutils.h"
 #include "libavutil/timestamp.h"
+#include "libavutil/bprint.h"
 #include "libavformat/os_support.h"
-#include "libswresample/swresample.h"
 
 #include "libavformat/ffm.h" // not public API
 
 # include "libavfilter/avcodec.h"
 # include "libavfilter/avfilter.h"
 # include "libavfilter/avfiltergraph.h"
-# include "libavfilter/buffersink.h"
 # include "libavfilter/buffersrc.h"
+# include "libavfilter/buffersink.h"
 
 #if HAVE_SYS_RESOURCE_H
 #include <sys/types.h>
@@ -96,6 +97,8 @@
 #define VSYNC_VFR         2
 #define VSYNC_DROP        0xff
 
+#define SINKA
+
 const char program_name[] = "ffmpeg";
 const int program_birth_year = 2000;
 
@@ -167,11 +170,6 @@ static int print_stats = 1;
 static int debug_ts = 0;
 static int current_time;
 
-static uint8_t *audio_buf;
-static unsigned int allocated_audio_buf_size;
-static uint8_t *async_buf;
-static unsigned int allocated_async_buf_size;
-
 #define DEFAULT_PASS_LOGFILENAME_PREFIX "ffmpeg2pass"
 
 typedef struct InputFilter {
@@ -229,8 +227,8 @@ typedef struct InputStream {
     /* dts of the last packet read for this stream */
     int64_t       dts;
 
-    int64_t       next_pts;  ///< synthetic pts for the next decode frame
-    int64_t       pts;       ///< current pts of the decoded frame
+    int64_t       next_pts;  /* synthetic pts for the next decode frame */
+    int64_t       pts;       /* current pts of the decoded frame */
     double ts_scale;
     int is_start;            /* is 1 at the start and after a discontinuity */
     int saw_first_ts;
@@ -241,12 +239,17 @@ typedef struct InputStream {
     int resample_width;
     int resample_pix_fmt;
 
+    int      resample_sample_fmt;
+    int      resample_sample_rate;
+    int      resample_channels;
+    uint64_t resample_channel_layout;
+
     /* a pool of free buffers for decoded data */
     FrameBuffer *buffer_pool;
     int dr1;
 
     /* decoded data from this stream goes into all those filters
-     * currently video only */
+     * currently video and audio only */
     InputFilter **filters;
     int        nb_filters;
 } InputStream;
@@ -273,10 +276,12 @@ typedef struct OutputStream {
        for A/V sync */
     struct InputStream *sync_ist; /* input stream to sync against */
     int64_t sync_opts;       /* output frame counter, could be changed to some true timestamp */ // FIXME look at frame_number
+    /* pts of the first frame encoded for this stream, used for limiting
+     * recording time */
+    int64_t first_pts;
     AVBitStreamFilterContext *bitstream_filters;
     AVCodec *enc;
     int64_t max_frames;
-    AVFrame *output_frame;
     AVFrame *filtered_frame;
 
     /* video only */
@@ -293,18 +298,10 @@ typedef struct OutputStream {
     int forced_kf_index;
 
     /* audio only */
-    int audio_resample;
-    int audio_channels_map[SWR_CH_MAX];  ///< list of the channels id to pick from the source stream
-    int audio_channels_mapped;           ///< number of channels in audio_channels_map
-    int resample_sample_fmt;
-    int resample_channels;
-    uint64_t resample_channel_layout;
-    int resample_sample_rate;
-    float rematrix_volume;
-    AVFifoBuffer *fifo;     /* for compression: one audio fifo per codec */
-    FILE *logfile;
+    int audio_channels_map[SWR_CH_MAX];  /* list of the channels id to pick from the source stream */
+    int audio_channels_mapped;           /* number of channels in audio_channels_map */
 
-    SwrContext *swr;
+    FILE *logfile;
 
     OutputFilter *filter;
     char *avfilter;
@@ -318,7 +315,7 @@ typedef struct OutputStream {
     const char *attachment_filename;
     int copy_initial_nonkeyframes;
 
-    enum PixelFormat pix_fmts[2];
+    int keep_pix_fmt;
 } OutputStream;
 
 
@@ -362,8 +359,6 @@ typedef struct OptionsContext {
     int        nb_audio_channels;
     SpecifierOpt *audio_sample_rate;
     int        nb_audio_sample_rate;
-    SpecifierOpt *rematrix_volume;
-    int        nb_rematrix_volume;
     SpecifierOpt *frame_rates;
     int        nb_frame_rates;
     SpecifierOpt *frame_sizes;
@@ -383,8 +378,8 @@ typedef struct OptionsContext {
     /* output options */
     StreamMap *stream_maps;
     int     nb_stream_maps;
-    AudioChannelMap *audio_channel_maps; ///< one info entry per -map_channel
-    int           nb_audio_channel_maps; ///< number of (valid) -map_channel settings
+    AudioChannelMap *audio_channel_maps; /* one info entry per -map_channel */
+    int           nb_audio_channel_maps; /* number of (valid) -map_channel settings */
     int metadata_global_manual;
     int metadata_streams_manual;
     int metadata_chapters_manual;
@@ -443,6 +438,8 @@ typedef struct OptionsContext {
     int        nb_filters;
 } OptionsContext;
 
+static void do_video_stats(AVFormatContext *os, OutputStream *ost, int frame_size);
+
 #define MATCH_PER_STREAM_OPT(name, type, outvar, fmtctx, st)\
 {\
     int i, ret;\
@@ -704,11 +701,19 @@ static enum PixelFormat choose_pixel_fmt(AVStream *st, AVCodec *codec, enum Pixe
     return target;
 }
 
-static char *choose_pixel_fmts(OutputStream *ost)
+static char *choose_pix_fmts(OutputStream *ost)
 {
+     if (ost->keep_pix_fmt) {
+        if (ost->filter)
+            avfilter_graph_set_auto_convert(ost->filter->graph->graph,
+                                            AVFILTER_AUTO_CONVERT_NONE);
+        if (ost->st->codec->pix_fmt == PIX_FMT_NONE)
+            return NULL;
+        return av_strdup(av_get_pix_fmt_name(ost->st->codec->pix_fmt));
+    }
     if (ost->st->codec->pix_fmt != PIX_FMT_NONE) {
         return av_strdup(av_get_pix_fmt_name(choose_pixel_fmt(ost->st, ost->enc, ost->st->codec->pix_fmt)));
-    } else if (ost->enc->pix_fmts) {
+    } else if (ost->enc && ost->enc->pix_fmts) {
         const enum PixelFormat *p;
         AVIOContext *s = NULL;
         uint8_t *ret;
@@ -727,8 +732,10 @@ static char *choose_pixel_fmts(OutputStream *ost)
             }
         }
 
-        for (; *p != PIX_FMT_NONE; p++)
-            avio_printf(s, "%s:", av_get_pix_fmt_name(*p));
+        for (; *p != PIX_FMT_NONE; p++) {
+            const char *name = av_get_pix_fmt_name(*p);
+            avio_printf(s, "%s:", name);
+        }
         len = avio_close_dyn_buf(s, &ret);
         ret[len - 1] = 0;
         return ret;
@@ -736,11 +743,208 @@ static char *choose_pixel_fmts(OutputStream *ost)
         return NULL;
 }
 
-static int configure_video_filters(FilterGraph *fg)
+/**
+ * Define a function for building a string containing a list of
+ * allowed formats,
+ */
+#define DEF_CHOOSE_FORMAT(type, var, supported_list, none, get_name, separator) \
+static char *choose_ ## var ## s(OutputStream *ost)                             \
+{                                                                               \
+    if (ost->st->codec->var != none) {                                          \
+        get_name(ost->st->codec->var);                                          \
+        return av_strdup(name);                                                 \
+    } else if (ost->enc->supported_list) {                                      \
+        const type *p;                                                          \
+        AVIOContext *s = NULL;                                                  \
+        uint8_t *ret;                                                           \
+        int len;                                                                \
+                                                                                \
+        if (avio_open_dyn_buf(&s) < 0)                                          \
+            exit_program(1);                                                    \
+                                                                                \
+        for (p = ost->enc->supported_list; *p != none; p++) {                   \
+            get_name(*p);                                                       \
+            avio_printf(s, "%s" separator, name);                               \
+        }                                                                       \
+        len = avio_close_dyn_buf(s, &ret);                                      \
+        ret[len - 1] = 0;                                                       \
+        return ret;                                                             \
+    } else                                                                      \
+        return NULL;                                                            \
+}
+
+#define GET_PIX_FMT_NAME(pix_fmt)\
+    const char *name = av_get_pix_fmt_name(pix_fmt);
+
+// DEF_CHOOSE_FORMAT(enum PixelFormat, pix_fmt, pix_fmts, PIX_FMT_NONE,
+//                   GET_PIX_FMT_NAME, ":")
+
+#define GET_SAMPLE_FMT_NAME(sample_fmt)\
+    const char *name = av_get_sample_fmt_name(sample_fmt)
+
+DEF_CHOOSE_FORMAT(enum AVSampleFormat, sample_fmt, sample_fmts,
+                  AV_SAMPLE_FMT_NONE, GET_SAMPLE_FMT_NAME, ",")
+
+#define GET_SAMPLE_RATE_NAME(rate)\
+    char name[16];\
+    snprintf(name, sizeof(name), "%d", rate);
+
+DEF_CHOOSE_FORMAT(int, sample_rate, supported_samplerates, 0,
+                  GET_SAMPLE_RATE_NAME, ",")
+
+#define GET_CH_LAYOUT_NAME(ch_layout)\
+    char name[16];\
+    snprintf(name, sizeof(name), "0x%"PRIx64, ch_layout);
+
+DEF_CHOOSE_FORMAT(uint64_t, channel_layout, channel_layouts, 0,
+                  GET_CH_LAYOUT_NAME, ",")
+
+static int configure_audio_filters(FilterGraph *fg, AVFilterContext **in_filter,
+                                   AVFilterContext **out_filter)
 {
     InputStream  *ist = fg->inputs[0]->ist;
     OutputStream *ost = fg->outputs[0]->ost;
-    AVFilterContext *in_filter, *out_filter, *filter;
+    AVCodecContext *codec  = ost->st->codec;
+    AVCodecContext *icodec = ist->st->codec;
+    char *sample_fmts, *sample_rates, *channel_layouts;
+    char args[256];
+    int ret;
+
+    avfilter_graph_free(&fg->graph);
+    if (!(fg->graph = avfilter_graph_alloc()))
+        return AVERROR(ENOMEM);
+
+    snprintf(args, sizeof(args), "time_base=%d/%d:sample_rate=%d:sample_fmt=%s:"
+             "channel_layout=0x%"PRIx64, ist->st->time_base.num,
+             ist->st->time_base.den, icodec->sample_rate,
+             av_get_sample_fmt_name(icodec->sample_fmt), icodec->channel_layout);
+    ret = avfilter_graph_create_filter(&fg->inputs[0]->filter,
+                                       avfilter_get_by_name("abuffer"),
+                                       "src", args, NULL, fg->graph);
+    if (ret < 0)
+        return ret;
+
+    ret = avfilter_graph_create_filter(&fg->outputs[0]->filter,
+                                       avfilter_get_by_name("abuffersink_old"),
+                                       "out", NULL, NULL, fg->graph);
+    if (ret < 0)
+        return ret;
+
+    *in_filter  = fg->inputs[0]->filter;
+    *out_filter = fg->outputs[0]->filter;
+
+    if (codec->channels && !codec->channel_layout)
+        codec->channel_layout = av_get_default_channel_layout(codec->channels);
+
+    sample_fmts     = choose_sample_fmts(ost);
+    sample_rates    = choose_sample_rates(ost);
+    channel_layouts = choose_channel_layouts(ost);
+    if (sample_fmts || sample_rates || channel_layouts) {
+        AVFilterContext *format;
+        char args[256];
+        int len = 0;
+
+        if (sample_fmts)
+            len += snprintf(args + len, sizeof(args) - len, "sample_fmts=%s:",
+                            sample_fmts);
+        if (sample_rates)
+            len += snprintf(args + len, sizeof(args) - len, "sample_rates=%s:",
+                            sample_rates);
+        if (channel_layouts)
+            len += snprintf(args + len, sizeof(args) - len, "channel_layouts=%s:",
+                            channel_layouts);
+        args[len - 1] = 0;
+
+        av_freep(&sample_fmts);
+        av_freep(&sample_rates);
+        av_freep(&channel_layouts);
+
+        ret = avfilter_graph_create_filter(&format,
+                                           avfilter_get_by_name("aformat"),
+                                           "aformat", args, NULL, fg->graph);
+        if (ret < 0)
+            return ret;
+
+        ret = avfilter_link(format, 0, fg->outputs[0]->filter, 0);
+        if (ret < 0)
+            return ret;
+
+        *out_filter = format;
+    }
+
+    if (audio_sync_method > 0) {
+        AVFilterContext *aswr;
+        char args[256] = {0};
+
+        av_strlcatf(args, sizeof(args), "min_comp=0.001:min_hard_comp=%f", audio_drift_threshold);
+
+        if (audio_sync_method > 1)
+            av_strlcatf(args, sizeof(args), ":max_soft_comp=%f", audio_sync_method/(double)icodec->sample_rate);
+
+        av_log(NULL, AV_LOG_INFO, "-async %d is forwarded to lavfi similarly to -af aresample=%s\n", audio_sync_method, args);
+
+        ret = avfilter_graph_create_filter(&aswr, avfilter_get_by_name("aresample"),
+                                           "aresample", args, NULL, fg->graph);
+        if (ret < 0)
+            return ret;
+
+        ret = avfilter_link(*in_filter, 0, aswr, 0);
+        if (ret < 0)
+            return ret;
+
+        *in_filter = aswr;
+    }
+
+#define AUTO_INSERT_FILTER(opt_name, filter_name, arg) do {                 \
+    AVFilterContext *filt_ctx;                                              \
+                                                                            \
+    av_log(NULL, AV_LOG_INFO, opt_name " is forwarded to lavfi "            \
+           "similarly to -af " filter_name "=%s.\n", arg);                  \
+                                                                            \
+    ret = avfilter_graph_create_filter(&filt_ctx,                           \
+                                       avfilter_get_by_name(filter_name),   \
+                                       filter_name, arg, NULL, fg->graph);  \
+    if (ret < 0)                                                            \
+        return ret;                                                         \
+                                                                            \
+    ret = avfilter_link(*in_filter, 0, filt_ctx, 0);                        \
+    if (ret < 0)                                                            \
+        return ret;                                                         \
+                                                                            \
+    *in_filter = filt_ctx;                                                  \
+} while (0)
+
+    if (ost->audio_channels_mapped) {
+        int i;
+        AVBPrint pan_buf;
+
+        av_bprint_init(&pan_buf, 256, 8192);
+        av_bprintf(&pan_buf, "0x%"PRIx64,
+                   av_get_default_channel_layout(ost->audio_channels_mapped));
+        for (i = 0; i < ost->audio_channels_mapped; i++)
+            if (ost->audio_channels_map[i] != -1)
+                av_bprintf(&pan_buf, ":c%d=c%d", i, ost->audio_channels_map[i]);
+
+        AUTO_INSERT_FILTER("-map_channel", "pan", pan_buf.str);
+        av_bprint_finalize(&pan_buf, NULL);
+    }
+
+    if (audio_volume != 256) {
+        char args[256];
+
+        snprintf(args, sizeof(args), "%lf", audio_volume / 256.);
+        AUTO_INSERT_FILTER("-vol", "volume", args);
+    }
+
+    return 0;
+}
+
+static int configure_video_filters(FilterGraph *fg, AVFilterContext **in_filter,
+                                   AVFilterContext **out_filter)
+{
+    InputStream  *ist = fg->inputs[0]->ist;
+    OutputStream *ost = fg->outputs[0]->ost;
+    AVFilterContext *filter;
     AVCodecContext *codec = ost->st->codec;
     AVBufferSinkParams *buffersink_params = av_buffersink_params_alloc();
     char *pix_fmts;
@@ -748,19 +952,16 @@ static int configure_video_filters(FilterGraph *fg)
     char args[255];
     int ret;
 
-    avfilter_graph_free(&fg->graph);
-    fg->graph = avfilter_graph_alloc();
-    if (!fg->graph)
-        return AVERROR(ENOMEM);
-
     if (ist->st->sample_aspect_ratio.num) {
         sample_aspect_ratio = ist->st->sample_aspect_ratio;
     } else
         sample_aspect_ratio = ist->st->codec->sample_aspect_ratio;
 
     snprintf(args, 255, "%d:%d:%d:%d:%d:%d:%d:flags=%d", ist->st->codec->width,
-             ist->st->codec->height, ist->st->codec->pix_fmt, 1, AV_TIME_BASE,
-             sample_aspect_ratio.num, sample_aspect_ratio.den, SWS_BILINEAR + ((ist->st->codec->flags&CODEC_FLAG_BITEXACT) ? SWS_BITEXACT:0));
+             ist->st->codec->height, ist->st->codec->pix_fmt,
+             ist->st->time_base.num, ist->st->time_base.den,
+             sample_aspect_ratio.num, sample_aspect_ratio.den,
+             SWS_BILINEAR + ((ist->st->codec->flags&CODEC_FLAG_BITEXACT) ? SWS_BITEXACT:0));
 
     ret = avfilter_graph_create_filter(&fg->inputs[0]->filter,
                                        avfilter_get_by_name("buffer"),
@@ -781,8 +982,8 @@ static int configure_video_filters(FilterGraph *fg)
 
     if (ret < 0)
         return ret;
-    in_filter  = fg->inputs[0]->filter;
-    out_filter = fg->outputs[0]->filter;
+    *in_filter  = fg->inputs[0]->filter;
+    *out_filter = fg->outputs[0]->filter;
 
     if (codec->width || codec->height) {
         snprintf(args, 255, "%d:%d:flags=0x%X",
@@ -792,27 +993,53 @@ static int configure_video_filters(FilterGraph *fg)
         if ((ret = avfilter_graph_create_filter(&filter, avfilter_get_by_name("scale"),
                                                 NULL, args, NULL, fg->graph)) < 0)
             return ret;
-        if ((ret = avfilter_link(in_filter, 0, filter, 0)) < 0)
+        if ((ret = avfilter_link(*in_filter, 0, filter, 0)) < 0)
             return ret;
-        in_filter = filter;
+        *in_filter = filter;
     }
 
-    if ((pix_fmts = choose_pixel_fmts(ost))) {
+    if ((pix_fmts = choose_pix_fmts(ost))) {
         if ((ret = avfilter_graph_create_filter(&filter,
                                                 avfilter_get_by_name("format"),
                                                 "format", pix_fmts, NULL,
                                                 fg->graph)) < 0)
             return ret;
-        if ((ret = avfilter_link(filter, 0, out_filter, 0)) < 0)
+        if ((ret = avfilter_link(filter, 0, *out_filter, 0)) < 0)
             return ret;
 
-        out_filter = filter;
+        *out_filter = filter;
         av_freep(&pix_fmts);
     }
 
     snprintf(args, sizeof(args), "flags=0x%X", (unsigned)ost->sws_flags);
     fg->graph->scale_sws_opts = av_strdup(args);
 
+    return 0;
+}
+
+static int configure_simple_filtergraph(FilterGraph *fg)
+{
+    OutputStream *ost = fg->outputs[0]->ost;
+    AVFilterContext *in_filter, *out_filter;
+    int ret;
+
+    avfilter_graph_free(&fg->graph);
+    fg->graph = avfilter_graph_alloc();
+    if (!fg->graph)
+        return AVERROR(ENOMEM);
+
+    switch (ost->st->codec->codec_type) {
+    case AVMEDIA_TYPE_VIDEO:
+        ret = configure_video_filters(fg, &in_filter, &out_filter);
+        break;
+    case AVMEDIA_TYPE_AUDIO:
+        ret = configure_audio_filters(fg, &in_filter, &out_filter);
+        break;
+    default: av_assert0(0);
+    }
+    if (ret < 0)
+        return ret;
+
     if (ost->avfilter) {
         AVFilterInOut *outputs = avfilter_inout_alloc();
         AVFilterInOut *inputs  = avfilter_inout_alloc();
@@ -835,6 +1062,10 @@ static int configure_video_filters(FilterGraph *fg)
             return ret;
     }
 
+    if (ost->keep_pix_fmt)
+        avfilter_graph_set_auto_convert(fg->graph,
+                                        AVFILTER_AUTO_CONVERT_NONE);
+
     if ((ret = avfilter_graph_config(fg->graph, NULL)) < 0)
         return ret;
 
@@ -883,8 +1114,9 @@ static void init_input_filter(FilterGraph *fg, AVFilterInOut *in)
     int i;
 
     // TODO: support other filter types
-    if (type != AVMEDIA_TYPE_VIDEO) {
-        av_log(NULL, AV_LOG_FATAL, "Only video filters supported currently.\n");
+    if (type != AVMEDIA_TYPE_VIDEO && type != AVMEDIA_TYPE_AUDIO) {
+        av_log(NULL, AV_LOG_FATAL, "Only video and audio filters supported "
+               "currently.\n");
         exit_program(1);
     }
 
@@ -945,7 +1177,7 @@ static void init_input_filter(FilterGraph *fg, AVFilterInOut *in)
     ist->filters[ist->nb_filters - 1] = fg->inputs[fg->nb_inputs - 1];
 }
 
-static int configure_output_filter(FilterGraph *fg, OutputFilter *ofilter, AVFilterInOut *out)
+static int configure_output_video_filter(FilterGraph *fg, OutputFilter *ofilter, AVFilterInOut *out)
 {
     char *pix_fmts;
     AVCodecContext *codec = ofilter->ost->st->codec;
@@ -986,7 +1218,7 @@ static int configure_output_filter(FilterGraph *fg, OutputFilter *ofilter, AVFil
         pad_idx = 0;
     }
 
-    if ((pix_fmts = choose_pixel_fmts(ofilter->ost))) {
+    if ((pix_fmts = choose_pix_fmts(ofilter->ost))) {
         AVFilterContext *filter;
         if ((ret = avfilter_graph_create_filter(&filter,
                                                 avfilter_get_by_name("format"),
@@ -1007,6 +1239,104 @@ static int configure_output_filter(FilterGraph *fg, OutputFilter *ofilter, AVFil
     return 0;
 }
 
+static int configure_output_audio_filter(FilterGraph *fg, OutputFilter *ofilter, AVFilterInOut *out)
+{
+    OutputStream *ost = ofilter->ost;
+    AVCodecContext *codec  = ost->st->codec;
+    AVFilterContext *last_filter = out->filter_ctx;
+    int pad_idx = out->pad_idx;
+    char *sample_fmts, *sample_rates, *channel_layouts;
+    int ret;
+
+    ret = avfilter_graph_create_filter(&ofilter->filter,
+                                       avfilter_get_by_name("abuffersink"),
+                                       "out", NULL, NULL, fg->graph);
+    if (ret < 0)
+        return ret;
+
+    if (codec->channels && !codec->channel_layout)
+        codec->channel_layout = av_get_default_channel_layout(codec->channels);
+
+    sample_fmts     = choose_sample_fmts(ost);
+    sample_rates    = choose_sample_rates(ost);
+    channel_layouts = choose_channel_layouts(ost);
+    if (sample_fmts || sample_rates || channel_layouts) {
+        AVFilterContext *format;
+        char args[256];
+        int len = 0;
+
+        if (sample_fmts)
+            len += snprintf(args + len, sizeof(args) - len, "sample_fmts=%s:",
+                            sample_fmts);
+        if (sample_rates)
+            len += snprintf(args + len, sizeof(args) - len, "sample_rates=%s:",
+                            sample_rates);
+        if (channel_layouts)
+            len += snprintf(args + len, sizeof(args) - len, "channel_layouts=%s:",
+                            channel_layouts);
+        args[len - 1] = 0;
+
+        av_freep(&sample_fmts);
+        av_freep(&sample_rates);
+        av_freep(&channel_layouts);
+
+        ret = avfilter_graph_create_filter(&format,
+                                           avfilter_get_by_name("aformat"),
+                                           "aformat", args, NULL, fg->graph);
+        if (ret < 0)
+            return ret;
+
+        ret = avfilter_link(last_filter, pad_idx, format, 0);
+        if (ret < 0)
+            return ret;
+
+        last_filter = format;
+        pad_idx = 0;
+    }
+
+    if (audio_sync_method > 0) {
+        AVFilterContext *async;
+        char args[256];
+        int  len = 0;
+
+        av_log(NULL, AV_LOG_WARNING, "-async has been deprecated. Used the "
+               "asyncts audio filter instead.\n");
+
+        if (audio_sync_method > 1)
+            len += snprintf(args + len, sizeof(args) - len, "compensate=1:"
+                            "max_comp=%d:", audio_sync_method);
+        snprintf(args + len, sizeof(args) - len, "min_delta=%f",
+                 audio_drift_threshold);
+
+        ret = avfilter_graph_create_filter(&async,
+                                           avfilter_get_by_name("asyncts"),
+                                           "async", args, NULL, fg->graph);
+        if (ret < 0)
+            return ret;
+
+        ret = avfilter_link(last_filter, pad_idx, async, 0);
+        if (ret < 0)
+            return ret;
+
+        last_filter = async;
+        pad_idx = 0;
+    }
+
+    if ((ret = avfilter_link(last_filter, pad_idx, ofilter->filter, 0)) < 0)
+        return ret;
+
+    return 0;
+}
+
+static int configure_output_filter(FilterGraph *fg, OutputFilter *ofilter, AVFilterInOut *out)
+{
+    switch (out->filter_ctx->output_pads[out->pad_idx].type) {
+    case AVMEDIA_TYPE_VIDEO: return configure_output_video_filter(fg, ofilter, out);
+    case AVMEDIA_TYPE_AUDIO: return configure_output_audio_filter(fg, ofilter, out);
+    default: av_assert0(0);
+    }
+}
+
 static int configure_complex_filter(FilterGraph *fg)
 {
     AVFilterInOut *inputs, *outputs, *cur;
@@ -1026,16 +1356,35 @@ static int configure_complex_filter(FilterGraph *fg)
         InputFilter *ifilter = fg->inputs[i];
         InputStream     *ist = ifilter->ist;
         AVRational       sar;
+        AVFilter     *filter;
         char            args[255];
 
-        sar = ist->st->sample_aspect_ratio.num ? ist->st->sample_aspect_ratio :
-                                                 ist->st->codec->sample_aspect_ratio;
-        snprintf(args, sizeof(args), "%d:%d:%d:%d:%d:%d:%d", ist->st->codec->width,
-                 ist->st->codec->height, ist->st->codec->pix_fmt, 1, AV_TIME_BASE,
-                 sar.num, sar.den);
+        switch (cur->filter_ctx->input_pads[cur->pad_idx].type) {
+        case AVMEDIA_TYPE_VIDEO:
+            sar = ist->st->sample_aspect_ratio.num ?
+                  ist->st->sample_aspect_ratio :
+                  ist->st->codec->sample_aspect_ratio;
+            snprintf(args, sizeof(args), "%d:%d:%d:%d:%d:%d:%d", ist->st->codec->width,
+                     ist->st->codec->height, ist->st->codec->pix_fmt,
+                     ist->st->time_base.num, ist->st->time_base.den,
+                     sar.num, sar.den);
+            filter = avfilter_get_by_name("buffer");
+            break;
+        case AVMEDIA_TYPE_AUDIO:
+            snprintf(args, sizeof(args), "time_base=%d/%d:sample_rate=%d:"
+                     "sample_fmt=%s:channel_layout=0x%"PRIx64,
+                     ist->st->time_base.num, ist->st->time_base.den,
+                     ist->st->codec->sample_rate,
+                     av_get_sample_fmt_name(ist->st->codec->sample_fmt),
+                     ist->st->codec->channel_layout);
+            filter = avfilter_get_by_name("abuffer");
+            break;
+        default:
+            av_assert0(0);
+        }
 
         if ((ret = avfilter_graph_create_filter(&ifilter->filter,
-                                                avfilter_get_by_name("buffer"), cur->name,
+                                                filter, cur->name,
                                                 args, NULL, fg->graph)) < 0)
             return ret;
         if ((ret = avfilter_link(ifilter->filter, 0,
@@ -1083,7 +1432,8 @@ static int configure_complex_filters(void)
 
 static int configure_filtergraph(FilterGraph *fg)
 {
-    return fg->graph_desc ? configure_complex_filter(fg) : configure_video_filters(fg);
+    return fg->graph_desc ? configure_complex_filter(fg) :
+                            configure_simple_filtergraph(fg);
 }
 
 static int ist_in_filtergraph(FilterGraph *fg, InputStream *ist)
@@ -1244,12 +1594,6 @@ void av_noreturn exit_program(int ret)
         }
         output_streams[i]->bitstream_filters = NULL;
 
-        if (output_streams[i]->output_frame) {
-            AVFrame *frame = output_streams[i]->output_frame;
-            if (frame->extended_data != frame->data)
-                av_freep(&frame->extended_data);
-            av_freep(&frame);
-        }
         av_freep(&output_streams[i]->filtered_frame);
         av_freep(&output_streams[i]);
     }
@@ -1275,10 +1619,6 @@ void av_noreturn exit_program(int ret)
     av_freep(&output_files);
 
     uninit_opts();
-    av_freep(&audio_buf);
-    allocated_audio_buf_size = 0;
-    av_freep(&async_buf);
-    allocated_async_buf_size = 0;
 
     avfilter_uninit();
     avformat_network_deinit();
@@ -1289,7 +1629,7 @@ void av_noreturn exit_program(int ret)
         exit (255);
     }
 
-    exit(ret); /* not all OS-es handle main() return value */
+    exit(ret);
 }
 
 static void assert_avoptions(AVDictionary *m)
@@ -1340,44 +1680,6 @@ static void choose_sample_fmt(AVStream *st, AVCodec *codec)
     }
 }
 
-static void choose_sample_rate(AVStream *st, AVCodec *codec)
-{
-    if (codec && codec->supported_samplerates) {
-        const int *p  = codec->supported_samplerates;
-        int best      = 0;
-        int best_dist = INT_MAX;
-        for (; *p; p++) {
-            int dist = abs(st->codec->sample_rate - *p);
-            if (dist < best_dist) {
-                best_dist = dist;
-                best      = *p;
-            }
-        }
-        if (best_dist) {
-            int i;
-            const int *sample_rates = codec->supported_samplerates;
-            av_log(st->codec, AV_LOG_WARNING,
-                   "Requested sampling rate (%dHz) unsupported, using %dHz instead\n"
-                   "Available sampling rates for %s:",
-                   st->codec->sample_rate, best, codec->name);
-            for (i = 0; sample_rates[i]; i++) {
-                if (!sample_rates[i + 1]) av_log(st->codec, AV_LOG_WARNING, " and");
-                else if (i)               av_log(st->codec, AV_LOG_WARNING, ",");
-                av_log(st->codec, AV_LOG_WARNING, " %d", sample_rates[i]);
-            }
-            av_log(st->codec, AV_LOG_WARNING, ".\n");
-        }
-        st->codec->sample_rate = best;
-    }
-}
-
-static double
-get_sync_ipts(const OutputStream *ost, int64_t pts)
-{
-    OutputFile *of = output_files[ost->file_index];
-    return (double)(pts - of->start_time) / AV_TIME_BASE;
-}
-
 static void write_frame(AVFormatContext *s, AVPacket *pkt, OutputStream *ost)
 {
     AVBitStreamFilterContext *bsfc = ost->bitstream_filters;
@@ -1388,6 +1690,14 @@ static void write_frame(AVFormatContext *s, AVPacket *pkt, OutputStream *ost)
         (avctx->codec_type == AVMEDIA_TYPE_AUDIO && audio_sync_method < 0))
         pkt->pts = pkt->dts = AV_NOPTS_VALUE;
 
+    if (avctx->codec_type == AVMEDIA_TYPE_AUDIO && pkt->dts != AV_NOPTS_VALUE) {
+        int64_t max = ost->st->cur_dts + !(s->oformat->flags & AVFMT_TS_NONSTRICT);
+        if (ost->st->cur_dts && ost->st->cur_dts != AV_NOPTS_VALUE &&  max > pkt->dts) {
+            av_log(s, max - pkt->dts > 2 ? AV_LOG_WARNING : AV_LOG_DEBUG, "Audio timestamp %"PRId64" < %"PRId64" invalid, cliping\n", pkt->dts, max);
+            pkt->pts = pkt->dts = max;
+        }
+    }
+
     /*
      * Audio encoders may split the packets --  #frames in != #packets out.
      * But there is no reordering, so we can limit the number of output packets
@@ -1433,104 +1743,38 @@ static void write_frame(AVFormatContext *s, AVPacket *pkt, OutputStream *ost)
     }
 }
 
-static void get_default_channel_layouts(OutputStream *ost, InputStream *ist)
+static int check_recording_time(OutputStream *ost)
 {
-    char layout_name[256];
-    AVCodecContext *enc = ost->st->codec;
-    AVCodecContext *dec = ist->st->codec;
-
-    if (dec->channel_layout &&
-        av_get_channel_layout_nb_channels(dec->channel_layout) != dec->channels) {
-        av_get_channel_layout_string(layout_name, sizeof(layout_name),
-                                     dec->channels, dec->channel_layout);
-        av_log(NULL, AV_LOG_ERROR, "New channel layout (%s) is invalid\n",
-               layout_name);
-        dec->channel_layout = 0;
-    }
-    if (!dec->channel_layout) {
-        if (enc->channel_layout && dec->channels == enc->channels) {
-            dec->channel_layout = enc->channel_layout;
-        } else {
-            dec->channel_layout = av_get_default_channel_layout(dec->channels);
+    OutputFile *of = output_files[ost->file_index];
 
-            if (!dec->channel_layout) {
-                av_log(NULL, AV_LOG_FATAL, "Unable to find default channel "
-                       "layout for Input Stream #%d.%d\n", ist->file_index,
-                       ist->st->index);
-                exit_program(1);
-            }
-        }
-        av_get_channel_layout_string(layout_name, sizeof(layout_name),
-                                     dec->channels, dec->channel_layout);
-        av_log(NULL, AV_LOG_WARNING, "Guessed Channel Layout for  Input Stream "
-               "#%d.%d : %s\n", ist->file_index, ist->st->index, layout_name);
-    }
-    if (!enc->channel_layout) {
-        if (dec->channels == enc->channels) {
-            enc->channel_layout = dec->channel_layout;
-            return;
-        } else {
-            enc->channel_layout = av_get_default_channel_layout(enc->channels);
-        }
-        if (!enc->channel_layout) {
-            av_log(NULL, AV_LOG_FATAL, "Unable to find default channel layout "
-                   "for Output Stream #%d.%d\n", ost->file_index,
-                   ost->st->index);
-            exit_program(1);
-        }
-        av_get_channel_layout_string(layout_name, sizeof(layout_name),
-                                     enc->channels, enc->channel_layout);
-        av_log(NULL, AV_LOG_WARNING, "Guessed Channel Layout for Output Stream "
-               "#%d.%d : %s\n", ost->file_index, ost->st->index, layout_name);
+    if (of->recording_time != INT64_MAX &&
+        av_compare_ts(ost->sync_opts - ost->first_pts, ost->st->codec->time_base, of->recording_time,
+                      AV_TIME_BASE_Q) >= 0) {
+        ost->is_past_recording_time = 1;
+        return 0;
     }
+    return 1;
 }
 
-static void generate_silence(uint8_t* buf, enum AVSampleFormat sample_fmt, size_t size)
-{
-    int fill_char = 0x00;
-    if (sample_fmt == AV_SAMPLE_FMT_U8)
-        fill_char = 0x80;
-    memset(buf, fill_char, size);
-}
-
-static int encode_audio_frame(AVFormatContext *s, OutputStream *ost,
-                              const uint8_t *buf, int buf_size)
+static void do_audio_out(AVFormatContext *s, OutputStream *ost,
+                         AVFrame *frame)
 {
     AVCodecContext *enc = ost->st->codec;
-    AVFrame *frame = NULL;
     AVPacket pkt;
-    int ret, got_packet;
+    int got_packet = 0;
 
     av_init_packet(&pkt);
     pkt.data = NULL;
     pkt.size = 0;
-
-    if (buf && buf_size) {
-        if (!ost->output_frame) {
-            ost->output_frame = avcodec_alloc_frame();
-            if (!ost->output_frame) {
-                av_log(NULL, AV_LOG_FATAL, "out-of-memory in encode_audio_frame()\n");
-                exit_program(1);
-            }
-        }
-        frame = ost->output_frame;
-        if (frame->extended_data != frame->data)
-            av_freep(&frame->extended_data);
-        avcodec_get_frame_defaults(frame);
-
-        frame->nb_samples  = buf_size /
-                             (enc->channels * av_get_bytes_per_sample(enc->sample_fmt));
-        if ((ret = avcodec_fill_audio_frame(frame, enc->channels, enc->sample_fmt,
-                                            buf, buf_size, 1)) < 0) {
-            av_log(NULL, AV_LOG_FATAL, "Audio encoding failed (avcodec_fill_audio_frame)\n");
-            exit_program(1);
-        }
-
+#if 0
+    if (!check_recording_time(ost))
+        return;
+#endif
+    if (frame->pts == AV_NOPTS_VALUE || audio_sync_method < 0)
         frame->pts = ost->sync_opts;
-        ost->sync_opts += frame->nb_samples;
-    }
+    ost->sync_opts = frame->pts + frame->nb_samples;
 
-    got_packet = 0;
+    av_assert0(pkt.size || !pkt.data);
     update_benchmark(NULL);
     if (avcodec_encode_audio2(enc, &pkt, frame, &got_packet) < 0) {
         av_log(NULL, AV_LOG_FATAL, "Audio encoding failed (avcodec_encode_audio2)\n");
@@ -1538,259 +1782,25 @@ static int encode_audio_frame(AVFormatContext *s, OutputStream *ost,
     }
     update_benchmark("encode_audio %d.%d", ost->file_index, ost->index);
 
-    ret = pkt.size;
-
     if (got_packet) {
         if (pkt.pts != AV_NOPTS_VALUE)
             pkt.pts      = av_rescale_q(pkt.pts,      enc->time_base, ost->st->time_base);
-        if (pkt.dts != AV_NOPTS_VALUE) {
-            int64_t max = ost->st->cur_dts + !(s->oformat->flags & AVFMT_TS_NONSTRICT);
+        if (pkt.dts != AV_NOPTS_VALUE)
             pkt.dts      = av_rescale_q(pkt.dts,      enc->time_base, ost->st->time_base);
-            if (ost->st->cur_dts && ost->st->cur_dts != AV_NOPTS_VALUE &&  max > pkt.dts) {
-                av_log(s, max - pkt.dts > 2 ? AV_LOG_WARNING : AV_LOG_DEBUG, "Audio timestamp %"PRId64" < %"PRId64" invalid, cliping\n", pkt.dts, max);
-                pkt.pts = pkt.dts = max;
-            }
-        }
         if (pkt.duration > 0)
             pkt.duration = av_rescale_q(pkt.duration, enc->time_base, ost->st->time_base);
 
-        write_frame(s, &pkt, ost);
-
-        audio_size += pkt.size;
-
-        av_free_packet(&pkt);
-    }
-
-    if (debug_ts) {
-        av_log(NULL, AV_LOG_INFO, "encoder -> type:audio "
-               "pkt_pts:%s pkt_pts_time:%s pkt_dts:%s pkt_dts_time:%s\n",
-               av_ts2str(pkt.pts), av_ts2timestr(pkt.pts, &ost->st->time_base),
-               av_ts2str(pkt.dts), av_ts2timestr(pkt.dts, &ost->st->time_base));
-    }
-
-    return ret;
-}
-
-static int alloc_audio_output_buf(AVCodecContext *dec, AVCodecContext *enc,
-                                  int nb_samples, int *buf_linesize)
-{
-    int64_t audio_buf_samples;
-    int audio_buf_size;
-
-    /* calculate required number of samples to allocate */
-    audio_buf_samples = ((int64_t)nb_samples * enc->sample_rate + dec->sample_rate) /
-                        dec->sample_rate;
-    audio_buf_samples = 4 * audio_buf_samples + 10000; // safety factors for resampling
-    audio_buf_samples = FFMAX(audio_buf_samples, enc->frame_size);
-    if (audio_buf_samples > INT_MAX)
-        return AVERROR(EINVAL);
-
-    audio_buf_size = av_samples_get_buffer_size(buf_linesize, enc->channels,
-                                                audio_buf_samples,
-                                                enc->sample_fmt, 0);
-    if (audio_buf_size < 0)
-        return audio_buf_size;
-
-    av_fast_malloc(&audio_buf, &allocated_audio_buf_size, audio_buf_size);
-    if (!audio_buf)
-        return AVERROR(ENOMEM);
-
-    return 0;
-}
-
-static void do_audio_out(AVFormatContext *s, OutputStream *ost,
-                         InputStream *ist, AVFrame *decoded_frame)
-{
-    uint8_t *buftmp;
-    int64_t size_out;
-
-    int frame_bytes, resample_changed;
-    AVCodecContext *enc = ost->st->codec;
-    AVCodecContext *dec = ist->st->codec;
-    int osize = av_get_bytes_per_sample(enc->sample_fmt);
-    int isize = av_get_bytes_per_sample(dec->sample_fmt);
-    uint8_t *buf[AV_NUM_DATA_POINTERS];
-    int size     = decoded_frame->nb_samples * dec->channels * isize;
-    int planes   = av_sample_fmt_is_planar(dec->sample_fmt) ? dec->channels : 1;
-    int i;
-    int out_linesize = 0;
-
-    av_assert0(planes <= AV_NUM_DATA_POINTERS);
-
-    for(i=0; i<planes; i++)
-        buf[i]= decoded_frame->data[i];
-
-
-    get_default_channel_layouts(ost, ist);
-
-    if (alloc_audio_output_buf(dec, enc, decoded_frame->nb_samples, &out_linesize) < 0) {
-        av_log(NULL, AV_LOG_FATAL, "Error allocating audio buffer\n");
-        exit_program(1);
-    }
-
-    if (audio_sync_method > 1                      ||
-        enc->channels       != dec->channels       ||
-        enc->channel_layout != dec->channel_layout ||
-        enc->sample_rate    != dec->sample_rate    ||
-        dec->sample_fmt     != enc->sample_fmt)
-        ost->audio_resample = 1;
-
-    resample_changed = ost->resample_sample_fmt  != dec->sample_fmt ||
-                       ost->resample_channels    != dec->channels   ||
-                       ost->resample_channel_layout != dec->channel_layout ||
-                       ost->resample_sample_rate != dec->sample_rate;
-
-    if ((ost->audio_resample && !ost->swr) || resample_changed || ost->audio_channels_mapped) {
-
-        if (resample_changed) {
-            av_log(NULL, AV_LOG_INFO, "Input stream #%d:%d frame changed from rate:%d fmt:%s ch:%d chl:0x%"PRIx64" to rate:%d fmt:%s ch:%d chl:0x%"PRIx64"\n",
-                   ist->file_index, ist->st->index,
-                   ost->resample_sample_rate, av_get_sample_fmt_name(ost->resample_sample_fmt),
-                   ost->resample_channels, ost->resample_channel_layout,
-                   dec->sample_rate, av_get_sample_fmt_name(dec->sample_fmt),
-                   dec->channels, dec->channel_layout);
-            ost->resample_sample_fmt  = dec->sample_fmt;
-            ost->resample_channels    = dec->channels;
-            ost->resample_channel_layout = dec->channel_layout;
-            ost->resample_sample_rate = dec->sample_rate;
-            swr_free(&ost->swr);
-        }
-        /* if audio_sync_method is >1 the resampler is needed for audio drift compensation */
-        if (audio_sync_method <= 1 && !ost->audio_channels_mapped &&
-            ost->resample_sample_fmt  == enc->sample_fmt &&
-            ost->resample_channels    == enc->channels   &&
-            ost->resample_channel_layout == enc->channel_layout &&
-            ost->resample_sample_rate == enc->sample_rate) {
-            //ost->swr = NULL;
-            ost->audio_resample = 0;
-        } else {
-            ost->swr = swr_alloc_set_opts(ost->swr,
-                                          enc->channel_layout, enc->sample_fmt, enc->sample_rate,
-                                          dec->channel_layout, dec->sample_fmt, dec->sample_rate,
-                                          0, NULL);
-            av_opt_set_int(ost->swr, "dither_method", ost->swr_dither_method,0);
-            av_opt_set_double(ost->swr, "dither_scale", ost->swr_dither_scale,0);
-            if (ost->audio_channels_mapped)
-                swr_set_channel_mapping(ost->swr, ost->audio_channels_map);
-            av_opt_set_double(ost->swr, "rmvol", ost->rematrix_volume, 0);
-            if (ost->audio_channels_mapped) {
-                av_opt_set_int(ost->swr, "icl", av_get_default_channel_layout(ost->audio_channels_mapped), 0);
-                av_opt_set_int(ost->swr, "uch", ost->audio_channels_mapped, 0);
-            }
-            if (av_opt_set_int(ost->swr, "ich", dec->channels, 0) < 0) {
-                av_log(NULL, AV_LOG_FATAL, "Unsupported number of input channels\n");
-                exit_program(1);
-            }
-            if (av_opt_set_int(ost->swr, "och", enc->channels, 0) < 0) {
-                av_log(NULL, AV_LOG_FATAL, "Unsupported number of output channels\n");
-                exit_program(1);
-            }
-            if(audio_sync_method>1) av_opt_set_int(ost->swr, "flags", SWR_FLAG_RESAMPLE, 0);
-            if(ost->swr && swr_init(ost->swr) < 0){
-                av_log(NULL, AV_LOG_FATAL, "swr_init() failed\n");
-                swr_free(&ost->swr);
-            }
-
-            if (!ost->swr) {
-                av_log(NULL, AV_LOG_FATAL, "Can not resample %d channels @ %d Hz to %d channels @ %d Hz\n",
-                        dec->channels, dec->sample_rate,
-                        enc->channels, enc->sample_rate);
-                exit_program(1);
-            }
-        }
-    }
-
-    av_assert0(ost->audio_resample || dec->sample_fmt==enc->sample_fmt);
-
-    if (audio_sync_method > 0) {
-        double delta = get_sync_ipts(ost, ist->pts) * enc->sample_rate - ost->sync_opts -
-                       av_fifo_size(ost->fifo) / (enc->channels * osize);
-        int idelta = delta * dec->sample_rate / enc->sample_rate;
-        int byte_delta = idelta * isize * dec->channels;
-
-        // FIXME resample delay
-        if (fabs(delta) > 50) {
-            if (ist->is_start || fabs(delta) > audio_drift_threshold*enc->sample_rate) {
-                if (byte_delta < 0) {
-                    byte_delta = FFMAX(byte_delta, -size);
-                    size += byte_delta;
-                    for (i=0; i<planes; i++)
-                        buf[i]  -= byte_delta/planes;
-                    av_log(NULL, AV_LOG_VERBOSE, "discarding %d audio samples\n",
-                           -byte_delta / (isize * dec->channels));
-                    if (!size)
-                        return;
-                    ist->is_start = 0;
-                } else {
-                    av_fast_malloc(&async_buf, &allocated_async_buf_size,
-                                   byte_delta + size);
-                    if (!async_buf) {
-                        av_log(NULL, AV_LOG_FATAL, "Out of memory in do_audio_out\n");
-                        exit_program(1);
-                    }
-
-                    if (alloc_audio_output_buf(dec, enc, decoded_frame->nb_samples + idelta, &out_linesize) < 0) {
-                        av_log(NULL, AV_LOG_FATAL, "Error allocating audio buffer\n");
-                        exit_program(1);
-                    }
-                    ist->is_start = 0;
-
-                    for (i=0; i<planes; i++) {
-                        uint8_t *t = async_buf + i*((byte_delta + size)/planes);
-                        generate_silence(t, dec->sample_fmt, byte_delta/planes);
-                        memcpy(t + byte_delta/planes, buf[i], size/planes);
-                        buf[i] = t;
-                    }
-                    size += byte_delta;
-                    av_log(NULL, AV_LOG_VERBOSE, "adding %d audio samples of silence\n", idelta);
-                }
-            } else if (audio_sync_method > 1) {
-                int comp = av_clip(delta, -audio_sync_method, audio_sync_method);
-                av_log(NULL, AV_LOG_VERBOSE, "compensating audio timestamp drift:%f compensation:%d in:%d\n",
-                       delta, comp, enc->sample_rate);
-//                fprintf(stderr, "drift:%f len:%d opts:%"PRId64" ipts:%"PRId64" fifo:%d\n", delta, -1, ost->sync_opts, (int64_t)(get_sync_ipts(ost) * enc->sample_rate), av_fifo_size(ost->fifo)/(ost->st->codec->channels * 2));
-                swr_set_compensation(ost->swr, comp, enc->sample_rate);
-            }
-        }
-    } else if (audio_sync_method == 0)
-        ost->sync_opts = lrintf(get_sync_ipts(ost, ist->pts) * enc->sample_rate) -
-                                av_fifo_size(ost->fifo) / (enc->channels * osize); // FIXME wrong
-
-    if (ost->audio_resample || ost->audio_channels_mapped) {
-        buftmp = audio_buf;
-        size_out = swr_convert(ost->swr, (      uint8_t*[]){buftmp},
-                                      allocated_audio_buf_size / (enc->channels * osize),
-                                      (const uint8_t **)buf,
-                                      size / (dec->channels * isize));
-        if (size_out < 0) {
-            av_log(NULL, AV_LOG_FATAL, "swr_convert failed\n");
-            exit_program(1);
-        }
-        size_out = size_out * enc->channels * osize;
-    } else {
-        buftmp = buf[0];
-        size_out = size;
-    }
-
-    av_assert0(ost->audio_resample || dec->sample_fmt==enc->sample_fmt);
-
-    /* now encode as many frames as possible */
-    if (!(enc->codec->capabilities & CODEC_CAP_VARIABLE_FRAME_SIZE)) {
-        /* output resampled raw samples */
-        if (av_fifo_realloc2(ost->fifo, av_fifo_size(ost->fifo) + size_out) < 0) {
-            av_log(NULL, AV_LOG_FATAL, "av_fifo_realloc2() failed\n");
-            exit_program(1);
+        if (debug_ts) {
+            av_log(NULL, AV_LOG_INFO, "encoder -> type:audio "
+                   "pkt_pts:%s pkt_pts_time:%s pkt_dts:%s pkt_dts_time:%s\n",
+                   av_ts2str(pkt.pts), av_ts2timestr(pkt.pts, &ost->st->time_base),
+                   av_ts2str(pkt.dts), av_ts2timestr(pkt.dts, &ost->st->time_base));
         }
-        av_fifo_generic_write(ost->fifo, buftmp, size_out, NULL);
 
-        frame_bytes = enc->frame_size * osize * enc->channels;
-
-        while (av_fifo_size(ost->fifo) >= frame_bytes) {
-            av_fifo_generic_read(ost->fifo, audio_buf, frame_bytes, NULL);
-            encode_audio_frame(s, ost, audio_buf, frame_bytes);
-        }
-    } else {
-        encode_audio_frame(s, ost, buftmp, size_out);
+        write_frame(s, &pkt, ost);
+
+        audio_size += pkt.size;
+        av_free_packet(&pkt);
     }
 }
 
@@ -1897,54 +1907,15 @@ static void do_subtitle_out(AVFormatContext *s,
     }
 }
 
-static double psnr(double d)
-{
-    return -10.0 * log(d) / log(10.0);
-}
-
-static void do_video_stats(AVFormatContext *os, OutputStream *ost,
-                           int frame_size)
-{
-    AVCodecContext *enc;
-    int frame_number;
-    double ti1, bitrate, avg_bitrate;
-
-    /* this is executed just the first time do_video_stats is called */
-    if (!vstats_file) {
-        vstats_file = fopen(vstats_filename, "w");
-        if (!vstats_file) {
-            perror("fopen");
-            exit_program(1);
-        }
-    }
-
-    enc = ost->st->codec;
-    if (enc->codec_type == AVMEDIA_TYPE_VIDEO) {
-        frame_number = ost->frame_number;
-        fprintf(vstats_file, "frame= %5d q= %2.1f ", frame_number, enc->coded_frame->quality / (float)FF_QP2LAMBDA);
-        if (enc->flags&CODEC_FLAG_PSNR)
-            fprintf(vstats_file, "PSNR= %6.2f ", psnr(enc->coded_frame->error[0] / (enc->width * enc->height * 255.0 * 255.0)));
-
-        fprintf(vstats_file,"f_size= %6d ", frame_size);
-        /* compute pts value */
-        ti1 = ost->sync_opts * av_q2d(enc->time_base);
-        if (ti1 < 0.01)
-            ti1 = 0.01;
-
-        bitrate     = (frame_size * 8) / av_q2d(enc->time_base) / 1000.0;
-        avg_bitrate = (double)(video_size * 8) / ti1 / 1000.0;
-        fprintf(vstats_file, "s_size= %8.0fkB time= %0.3f br= %7.1fkbits/s avg_br= %7.1fkbits/s ",
-               (double)video_size / 1024, ti1, bitrate, avg_bitrate);
-        fprintf(vstats_file, "type= %c\n", av_get_picture_type_char(enc->coded_frame->pict_type));
-    }
-}
-
-
-static void do_video_out(AVFormatContext *s, OutputStream *ost,
-                         AVFrame *in_picture, float quality)
+static void do_video_out(AVFormatContext *s,
+                         OutputStream *ost,
+                         AVFrame *in_picture,
+                         float quality)
 {
-    int nb_frames, i, ret, format_video_sync;
-    AVCodecContext *enc;
+    int ret, format_video_sync;
+    AVPacket pkt;
+    AVCodecContext *enc = ost->st->codec;
+    int nb_frames, i;
     double sync_ipts, delta;
     double duration = 0;
     int frame_size = 0;
@@ -1953,12 +1924,10 @@ static void do_video_out(AVFormatContext *s, OutputStream *ost,
     if (ost->source_index >= 0)
         ist = input_streams[ost->source_index];
 
-    enc = ost->st->codec;
-
     if(ist && ist->st->start_time != AV_NOPTS_VALUE && ist->st->first_dts != AV_NOPTS_VALUE && ost->frame_rate.num)
         duration = 1/(av_q2d(ost->frame_rate) * av_q2d(enc->time_base));
 
-    sync_ipts = get_sync_ipts(ost, in_picture->pts) / av_q2d(enc->time_base);
+    sync_ipts = in_picture->pts;
     delta = sync_ipts - ost->sync_opts + duration;
 
     /* by default, we output a single frame */
@@ -2000,99 +1969,144 @@ static void do_video_out(AVFormatContext *s, OutputStream *ost,
         av_log(NULL, AV_LOG_VERBOSE, "*** %d dup!\n", nb_frames - 1);
     }
 
-    /* duplicates frame if needed */
-    for (i = 0; i < nb_frames; i++) {
-        AVPacket pkt;
-        av_init_packet(&pkt);
-        pkt.data = NULL;
-        pkt.size = 0;
-
-        if (s->oformat->flags & AVFMT_RAWPICTURE &&
-            enc->codec->id == CODEC_ID_RAWVIDEO) {
-            /* raw pictures are written as AVPicture structure to
-               avoid any copies. We support temporarily the older
-               method. */
-            enc->coded_frame->interlaced_frame = in_picture->interlaced_frame;
-            enc->coded_frame->top_field_first  = in_picture->top_field_first;
-            pkt.data   = (uint8_t *)in_picture;
-            pkt.size   =  sizeof(AVPicture);
-            pkt.pts    = av_rescale_q(ost->sync_opts, enc->time_base, ost->st->time_base);
-            pkt.flags |= AV_PKT_FLAG_KEY;
 
-            write_frame(s, &pkt, ost);
-        } else {
-            int got_packet;
-            AVFrame big_picture;
-
-            big_picture = *in_picture;
-            /* better than nothing: use input picture interlaced
-               settings */
-            big_picture.interlaced_frame = in_picture->interlaced_frame;
-            if (ost->st->codec->flags & (CODEC_FLAG_INTERLACED_DCT|CODEC_FLAG_INTERLACED_ME)) {
-                if (ost->top_field_first == -1)
-                    big_picture.top_field_first = in_picture->top_field_first;
-                else
-                    big_picture.top_field_first = !!ost->top_field_first;
-            }
+duplicate_frame:
+    av_init_packet(&pkt);
+    pkt.data = NULL;
+    pkt.size = 0;
 
-            /* handles same_quant here. This is not correct because it may
-               not be a global option */
-            big_picture.quality = quality;
-            if (!enc->me_threshold)
-                big_picture.pict_type = 0;
-            big_picture.pts = ost->sync_opts;
-            if (ost->forced_kf_index < ost->forced_kf_count &&
-                big_picture.pts >= ost->forced_kf_pts[ost->forced_kf_index]) {
-                big_picture.pict_type = AV_PICTURE_TYPE_I;
-                ost->forced_kf_index++;
-            }
-            update_benchmark(NULL);
-            ret = avcodec_encode_video2(enc, &pkt, &big_picture, &got_packet);
-            update_benchmark("encode_video %d.%d", ost->file_index, ost->index);
-            if (ret < 0) {
-                av_log(NULL, AV_LOG_FATAL, "Video encoding failed\n");
-                exit_program(1);
-            }
+    in_picture->pts = ost->sync_opts;
 
-            if (got_packet) {
-                if (pkt.pts == AV_NOPTS_VALUE && !(enc->codec->capabilities & CODEC_CAP_DELAY))
-                    pkt.pts = ost->sync_opts;
+    if (s->oformat->flags & AVFMT_RAWPICTURE &&
+        enc->codec->id == CODEC_ID_RAWVIDEO) {
+        /* raw pictures are written as AVPicture structure to
+           avoid any copies. We support temporarily the older
+           method. */
+        enc->coded_frame->interlaced_frame = in_picture->interlaced_frame;
+        enc->coded_frame->top_field_first  = in_picture->top_field_first;
+        pkt.data   = (uint8_t *)in_picture;
+        pkt.size   =  sizeof(AVPicture);
+        pkt.pts    = av_rescale_q(in_picture->pts, enc->time_base, ost->st->time_base);
+        pkt.flags |= AV_PKT_FLAG_KEY;
 
-                if (pkt.pts != AV_NOPTS_VALUE)
-                    pkt.pts = av_rescale_q(pkt.pts, enc->time_base, ost->st->time_base);
-                if (pkt.dts != AV_NOPTS_VALUE)
-                    pkt.dts = av_rescale_q(pkt.dts, enc->time_base, ost->st->time_base);
+        write_frame(s, &pkt, ost);
+    } else {
+        int got_packet;
+        AVFrame big_picture;
+
+        big_picture = *in_picture;
+        /* better than nothing: use input picture interlaced
+           settings */
+        big_picture.interlaced_frame = in_picture->interlaced_frame;
+        if (ost->st->codec->flags & (CODEC_FLAG_INTERLACED_DCT|CODEC_FLAG_INTERLACED_ME)) {
+            if (ost->top_field_first == -1)
+                big_picture.top_field_first = in_picture->top_field_first;
+            else
+                big_picture.top_field_first = !!ost->top_field_first;
+        }
+
+        /* handles same_quant here. This is not correct because it may
+           not be a global option */
+        big_picture.quality = quality;
+        if (!enc->me_threshold)
+            big_picture.pict_type = 0;
+        if (ost->forced_kf_index < ost->forced_kf_count &&
+            big_picture.pts >= ost->forced_kf_pts[ost->forced_kf_index]) {
+            big_picture.pict_type = AV_PICTURE_TYPE_I;
+            ost->forced_kf_index++;
+        }
+        update_benchmark(NULL);
+        ret = avcodec_encode_video2(enc, &pkt, &big_picture, &got_packet);
+        update_benchmark("encode_video %d.%d", ost->file_index, ost->index);
+        if (ret < 0) {
+            av_log(NULL, AV_LOG_FATAL, "Video encoding failed\n");
+            exit_program(1);
+        }
 
-                if (debug_ts) {
-                    av_log(NULL, AV_LOG_INFO, "encoder -> type:video "
-                           "pkt_pts:%s pkt_pts_time:%s pkt_dts:%s pkt_dts_time:%s\n",
-                           av_ts2str(pkt.pts), av_ts2timestr(pkt.pts, &ost->st->time_base),
-                           av_ts2str(pkt.dts), av_ts2timestr(pkt.dts, &ost->st->time_base));
-                }
+        if (got_packet) {
+            if (pkt.pts == AV_NOPTS_VALUE && !(enc->codec->capabilities & CODEC_CAP_DELAY))
+                pkt.pts = ost->sync_opts;
 
-                write_frame(s, &pkt, ost);
-                frame_size = pkt.size;
-                video_size += pkt.size;
-                av_free_packet(&pkt);
+            if (pkt.pts != AV_NOPTS_VALUE)
+                pkt.pts = av_rescale_q(pkt.pts, enc->time_base, ost->st->time_base);
+            if (pkt.dts != AV_NOPTS_VALUE)
+                pkt.dts = av_rescale_q(pkt.dts, enc->time_base, ost->st->time_base);
 
-                /* if two pass, output log */
-                if (ost->logfile && enc->stats_out) {
-                    fprintf(ost->logfile, "%s", enc->stats_out);
-                }
+            if (debug_ts) {
+                av_log(NULL, AV_LOG_INFO, "encoder -> type:video "
+                    "pkt_pts:%s pkt_pts_time:%s pkt_dts:%s pkt_dts_time:%s\n",
+                    av_ts2str(pkt.pts), av_ts2timestr(pkt.pts, &ost->st->time_base),
+                    av_ts2str(pkt.dts), av_ts2timestr(pkt.dts, &ost->st->time_base));
+            }
+
+            write_frame(s, &pkt, ost);
+            frame_size = pkt.size;
+            video_size += pkt.size;
+            av_free_packet(&pkt);
+
+            /* if two pass, output log */
+            if (ost->logfile && enc->stats_out) {
+                fprintf(ost->logfile, "%s", enc->stats_out);
             }
         }
-        ost->sync_opts++;
-        /*
-         * For video, number of frames in == number of packets out.
-         * But there may be reordering, so we can't throw away frames on encoder
-         * flush, we need to limit them here, before they go into encoder.
-         */
-        ost->frame_number++;
     }
+    ost->sync_opts++;
+    /*
+     * For video, number of frames in == number of packets out.
+     * But there may be reordering, so we can't throw away frames on encoder
+     * flush, we need to limit them here, before they go into encoder.
+     */
+    ost->frame_number++;
+
+    if(--nb_frames)
+        goto duplicate_frame;
+
     if (vstats_filename && frame_size)
         do_video_stats(output_files[ost->file_index]->ctx, ost, frame_size);
 }
 
+static double psnr(double d)
+{
+    return -10.0 * log(d) / log(10.0);
+}
+
+static void do_video_stats(AVFormatContext *os, OutputStream *ost,
+                           int frame_size)
+{
+    AVCodecContext *enc;
+    int frame_number;
+    double ti1, bitrate, avg_bitrate;
+
+    /* this is executed just the first time do_video_stats is called */
+    if (!vstats_file) {
+        vstats_file = fopen(vstats_filename, "w");
+        if (!vstats_file) {
+            perror("fopen");
+            exit_program(1);
+        }
+    }
+
+    enc = ost->st->codec;
+    if (enc->codec_type == AVMEDIA_TYPE_VIDEO) {
+        frame_number = ost->frame_number;
+        fprintf(vstats_file, "frame= %5d q= %2.1f ", frame_number, enc->coded_frame->quality / (float)FF_QP2LAMBDA);
+        if (enc->flags&CODEC_FLAG_PSNR)
+            fprintf(vstats_file, "PSNR= %6.2f ", psnr(enc->coded_frame->error[0] / (enc->width * enc->height * 255.0 * 255.0)));
+
+        fprintf(vstats_file,"f_size= %6d ", frame_size);
+        /* compute pts value */
+        ti1 = ost->sync_opts * av_q2d(enc->time_base);
+        if (ti1 < 0.01)
+            ti1 = 0.01;
+
+        bitrate     = (frame_size * 8) / av_q2d(enc->time_base) / 1000.0;
+        avg_bitrate = (double)(video_size * 8) / ti1 / 1000.0;
+        fprintf(vstats_file, "s_size= %8.0fkB time= %0.3f br= %7.1fkbits/s avg_br= %7.1fkbits/s ",
+               (double)video_size / 1024, ti1, bitrate, avg_bitrate);
+        fprintf(vstats_file, "type= %c\n", av_get_picture_type_char(enc->coded_frame->pict_type));
+    }
+}
+
 /* check for new output on any of the filtergraphs */
 static int poll_filters(void)
 {
@@ -2100,12 +2114,14 @@ static int poll_filters(void)
     AVFrame *filtered_frame = NULL;
     int i, ret, ret_all;
     unsigned nb_success, nb_eof;
+    int64_t frame_pts;
 
     while (1) {
         /* Reap all buffers present in the buffer sinks */
         for (i = 0; i < nb_output_streams; i++) {
             OutputStream *ost = output_streams[i];
             OutputFile    *of = output_files[ost->file_index];
+            int ret = 0;
 
             if (!ost->filter || ost->is_past_recording_time)
                 continue;
@@ -2118,10 +2134,19 @@ static int poll_filters(void)
 
             while (1) {
                 AVRational ist_pts_tb = ost->filter->filter->inputs[0]->time_base;
-                ret = av_buffersink_get_buffer_ref(ost->filter->filter, &picref,
-                                                   AV_BUFFERSINK_FLAG_NO_REQUEST);
+                if (ost->enc->type == AVMEDIA_TYPE_AUDIO &&
+                    !(ost->enc->capabilities & CODEC_CAP_VARIABLE_FRAME_SIZE))
+                    ret = av_buffersink_read_samples(ost->filter->filter, &picref,
+                                                    ost->st->codec->frame_size);
+                else
+#ifdef SINKA
+                    ret = av_buffersink_read(ost->filter->filter, &picref);
+#else
+                    ret = av_buffersink_get_buffer_ref(ost->filter->filter, &picref,
+                                                       AV_BUFFERSINK_FLAG_NO_REQUEST);
+#endif
                 if (ret < 0) {
-                    if (ret != AVERROR(EAGAIN)) {
+                    if (ret != AVERROR(EAGAIN) && ret != AVERROR_EOF) {
                         char buf[256];
                         av_strerror(ret, buf, sizeof(buf));
                         av_log(NULL, AV_LOG_WARNING,
@@ -2129,16 +2154,28 @@ static int poll_filters(void)
                     }
                     break;
                 }
-                filtered_frame->pts = av_rescale_q(picref->pts, ist_pts_tb, AV_TIME_BASE_Q);
+                frame_pts = AV_NOPTS_VALUE;
+                if (picref->pts != AV_NOPTS_VALUE) {
+                    filtered_frame->pts = frame_pts = av_rescale_q(picref->pts,
+                                                    ost->filter->filter->inputs[0]->time_base,
+                                                    ost->st->codec->time_base) -
+                                        av_rescale_q(of->start_time,
+                                                    AV_TIME_BASE_Q,
+                                                    ost->st->codec->time_base);
+
+                    if (of->start_time && filtered_frame->pts < 0) {
+                        avfilter_unref_buffer(picref);
+                        continue;
+                    }
+                }
                 //if (ost->source_index >= 0)
                 //    *filtered_frame= *input_streams[ost->source_index]->decoded_frame; //for me_threshold
 
-                if (of->start_time && filtered_frame->pts < of->start_time)
-                    return 0;
 
                 switch (ost->filter->filter->inputs[0]->type) {
                 case AVMEDIA_TYPE_VIDEO:
                     avfilter_fill_frame_from_video_buffer_ref(filtered_frame, picref);
+                    filtered_frame->pts = frame_pts;
                     if (!ost->frame_aspect_ratio)
                         ost->st->codec->sample_aspect_ratio = picref->video->sample_aspect_ratio;
 
@@ -2146,8 +2183,13 @@ static int poll_filters(void)
                                  same_quant ? ost->last_quality :
                                               ost->st->codec->global_quality);
                     break;
+                case AVMEDIA_TYPE_AUDIO:
+                    avfilter_copy_buf_props(filtered_frame, picref);
+                    filtered_frame->pts = frame_pts;
+                    do_audio_out(of->ctx, ost, filtered_frame);
+                    break;
                 default:
-                    // TODO support audio/subtitle filters
+                    // TODO support subtitle filters
                     av_assert0(0);
                 }
 
@@ -2330,40 +2372,40 @@ static void flush_encoders(void)
             continue;
 
         for (;;) {
-            AVPacket pkt;
-            int fifo_bytes, got_packet;
-            av_init_packet(&pkt);
-            pkt.data = NULL;
-            pkt.size = 0;
+            int (*encode)(AVCodecContext*, AVPacket*, const AVFrame*, int*) = NULL;
+            const char *desc;
+            int64_t *size;
 
             switch (ost->st->codec->codec_type) {
             case AVMEDIA_TYPE_AUDIO:
-                fifo_bytes = av_fifo_size(ost->fifo);
-                if (fifo_bytes > 0) {
-                    /* encode any samples remaining in fifo */
-                    int frame_bytes = fifo_bytes;
-
-                    av_fifo_generic_read(ost->fifo, audio_buf, fifo_bytes, NULL);
-
-                    encode_audio_frame(os, ost, audio_buf, frame_bytes);
-                } else {
-                    /* flush encoder with NULL frames until it is done
-                       returning packets */
-                    if (encode_audio_frame(os, ost, NULL, 0) == 0) {
-                        stop_encoding = 1;
-                        break;
-                    }
-                }
+                encode = avcodec_encode_audio2;
+                desc   = "Audio";
+                size   = &audio_size;
                 break;
             case AVMEDIA_TYPE_VIDEO:
+                encode = avcodec_encode_video2;
+                desc   = "Video";
+                size   = &video_size;
+                break;
+            default:
+                stop_encoding = 1;
+            }
+
+            if (encode) {
+                AVPacket pkt;
+                int got_packet;
+                av_init_packet(&pkt);
+                pkt.data = NULL;
+                pkt.size = 0;
+
                 update_benchmark(NULL);
-                ret = avcodec_encode_video2(enc, &pkt, NULL, &got_packet);
-                update_benchmark("encode_video %d.%d", ost->file_index, ost->index);
+                ret = encode(enc, &pkt, NULL, &got_packet);
+                update_benchmark("flush %s %d.%d", desc, ost->file_index, ost->index);
                 if (ret < 0) {
-                    av_log(NULL, AV_LOG_FATAL, "Video encoding failed\n");
+                    av_log(NULL, AV_LOG_FATAL, "%s encoding failed\n", desc);
                     exit_program(1);
                 }
-                video_size += pkt.size;
+                *size += pkt.size;
                 if (ost->logfile && enc->stats_out) {
                     fprintf(ost->logfile, "%s", enc->stats_out);
                 }
@@ -2376,10 +2418,8 @@ static void flush_encoders(void)
                 if (pkt.dts != AV_NOPTS_VALUE)
                     pkt.dts = av_rescale_q(pkt.dts, enc->time_base, ost->st->time_base);
                 write_frame(os, &pkt, ost);
-                break;
-            default:
-                stop_encoding = 1;
             }
+
             if (stop_encoding)
                 break;
         }
@@ -2480,12 +2520,29 @@ static void rate_emu_sleep(InputStream *ist)
     }
 }
 
-static int transcode_audio(InputStream *ist, AVPacket *pkt, int *got_output)
+static int guess_input_channel_layout(InputStream *ist)
+{
+    AVCodecContext *dec = ist->st->codec;
+
+    if (!dec->channel_layout) {
+        char layout_name[256];
+
+        dec->channel_layout = av_get_default_channel_layout(dec->channels);
+        if (!dec->channel_layout)
+            return 0;
+        av_get_channel_layout_string(layout_name, sizeof(layout_name),
+                                     dec->channels, dec->channel_layout);
+        av_log(NULL, AV_LOG_WARNING, "Guessed Channel Layout for  Input Stream "
+               "#%d.%d : %s\n", ist->file_index, ist->st->index, layout_name);
+    }
+    return 1;
+}
+
+static int decode_audio(InputStream *ist, AVPacket *pkt, int *got_output)
 {
     AVFrame *decoded_frame;
     AVCodecContext *avctx = ist->st->codec;
-    int bps = av_get_bytes_per_sample(ist->st->codec->sample_fmt);
-    int i, ret;
+    int i, ret, resample_changed;
 
     if (!ist->decoded_frame && !(ist->decoded_frame = avcodec_alloc_frame()))
         return AVERROR(ENOMEM);
@@ -2506,6 +2563,10 @@ static int transcode_audio(InputStream *ist, AVPacket *pkt, int *got_output)
 
     if (!*got_output) {
         /* no audio frame */
+        if (!pkt->size)
+            for (i = 0; i < ist->nb_filters; i++)
+                av_buffersrc_add_ref(ist->filters[i]->filter, NULL,
+                                     AV_BUFFERSRC_FLAG_NO_COPY);
         return ret;
     }
 
@@ -2513,87 +2574,72 @@ static int transcode_audio(InputStream *ist, AVPacket *pkt, int *got_output)
        the decoder could be delaying output by a packet or more. */
     if (decoded_frame->pts != AV_NOPTS_VALUE)
         ist->dts = ist->next_dts = ist->pts = ist->next_pts = decoded_frame->pts;
+    else if (pkt->pts != AV_NOPTS_VALUE) {
+        decoded_frame->pts = pkt->pts;
+        pkt->pts           = AV_NOPTS_VALUE;
+    }else
+        decoded_frame->pts = av_rescale_q(ist->dts, AV_TIME_BASE_Q, ist->st->time_base);
+
 
+#if 1
     /* increment next_dts to use for the case where the input stream does not
        have timestamps or there are multiple frames in the packet */
     ist->next_pts += ((int64_t)AV_TIME_BASE * decoded_frame->nb_samples) /
                      avctx->sample_rate;
     ist->next_dts += ((int64_t)AV_TIME_BASE * decoded_frame->nb_samples) /
                      avctx->sample_rate;
+#endif
 
+    rate_emu_sleep(ist);
 
-    // preprocess audio (volume)
-    if (audio_volume != 256) {
-        int decoded_data_size = decoded_frame->nb_samples * avctx->channels * bps;
-        void *samples = decoded_frame->data[0];
-        switch (avctx->sample_fmt) {
-        case AV_SAMPLE_FMT_U8:
-        {
-            uint8_t *volp = samples;
-            for (i = 0; i < (decoded_data_size / sizeof(*volp)); i++) {
-                int v = (((*volp - 128) * audio_volume + 128) >> 8) + 128;
-                *volp++ = av_clip_uint8(v);
-            }
-            break;
-        }
-        case AV_SAMPLE_FMT_S16:
-        {
-            int16_t *volp = samples;
-            for (i = 0; i < (decoded_data_size / sizeof(*volp)); i++) {
-                int v = ((*volp) * audio_volume + 128) >> 8;
-                *volp++ = av_clip_int16(v);
-            }
-            break;
-        }
-        case AV_SAMPLE_FMT_S32:
-        {
-            int32_t *volp = samples;
-            for (i = 0; i < (decoded_data_size / sizeof(*volp)); i++) {
-                int64_t v = (((int64_t)*volp * audio_volume + 128) >> 8);
-                *volp++ = av_clipl_int32(v);
-            }
-            break;
-        }
-        case AV_SAMPLE_FMT_FLT:
-        {
-            float *volp = samples;
-            float scale = audio_volume / 256.f;
-            for (i = 0; i < (decoded_data_size / sizeof(*volp)); i++) {
-                *volp++ *= scale;
-            }
-            break;
-        }
-        case AV_SAMPLE_FMT_DBL:
-        {
-            double *volp = samples;
-            double scale = audio_volume / 256.;
-            for (i = 0; i < (decoded_data_size / sizeof(*volp)); i++) {
-                *volp++ *= scale;
-            }
-            break;
-        }
-        default:
-            av_log(NULL, AV_LOG_FATAL,
-                   "Audio volume adjustment on sample format %s is not supported.\n",
-                   av_get_sample_fmt_name(ist->st->codec->sample_fmt));
+    resample_changed = ist->resample_sample_fmt     != decoded_frame->format         ||
+                       ist->resample_channels       != avctx->channels               ||
+                       ist->resample_channel_layout != decoded_frame->channel_layout ||
+                       ist->resample_sample_rate    != decoded_frame->sample_rate;
+    if (resample_changed) {
+        char layout1[64], layout2[64];
+
+        if (!guess_input_channel_layout(ist)) {
+            av_log(NULL, AV_LOG_FATAL, "Unable to find default channel "
+                   "layout for Input Stream #%d.%d\n", ist->file_index,
+                   ist->st->index);
             exit_program(1);
         }
-    }
+        decoded_frame->channel_layout = avctx->channel_layout;
 
-    rate_emu_sleep(ist);
+        av_get_channel_layout_string(layout1, sizeof(layout1), ist->resample_channels,
+                                     ist->resample_channel_layout);
+        av_get_channel_layout_string(layout2, sizeof(layout2), avctx->channels,
+                                     decoded_frame->channel_layout);
 
-    for (i = 0; i < nb_output_streams; i++) {
-        OutputStream *ost = output_streams[i];
+        av_log(NULL, AV_LOG_INFO,
+               "Input stream #%d:%d frame changed from rate:%d fmt:%s ch:%d chl:%s to rate:%d fmt:%s ch:%d chl:%s\n",
+               ist->file_index, ist->st->index,
+               ist->resample_sample_rate,  av_get_sample_fmt_name(ist->resample_sample_fmt),
+               ist->resample_channels, layout1,
+               decoded_frame->sample_rate, av_get_sample_fmt_name(decoded_frame->format),
+               avctx->channels, layout2);
 
-        if (!check_output_constraints(ist, ost) || !ost->encoding_needed)
-            continue;
-        do_audio_out(output_files[ost->file_index]->ctx, ost, ist, decoded_frame);
+        ist->resample_sample_fmt     = decoded_frame->format;
+        ist->resample_sample_rate    = decoded_frame->sample_rate;
+        ist->resample_channel_layout = decoded_frame->channel_layout;
+        ist->resample_channels       = avctx->channels;
+
+        for (i = 0; i < nb_filtergraphs; i++)
+            if (ist_in_filtergraph(filtergraphs[i], ist) &&
+                configure_filtergraph(filtergraphs[i]) < 0) {
+                av_log(NULL, AV_LOG_FATAL, "Error reinitializing filters!\n");
+                exit_program(1);
+            }
     }
 
+    for (i = 0; i < ist->nb_filters; i++)
+        av_buffersrc_add_frame(ist->filters[i]->filter, decoded_frame, 0);
+
     return ret;
 }
 
-static int transcode_video(InputStream *ist, AVPacket *pkt, int *got_output, int64_t *pkt_pts)
+static int decode_video(InputStream *ist, AVPacket *pkt, int *got_output)
 {
     AVFrame *decoded_frame;
     void *buffer_to_free = NULL;
@@ -2607,9 +2653,7 @@ static int transcode_video(InputStream *ist, AVPacket *pkt, int *got_output, int
     else
         avcodec_get_frame_defaults(ist->decoded_frame);
     decoded_frame = ist->decoded_frame;
-    pkt->pts  = *pkt_pts;
-    pkt->dts  = ist->dts;
-    *pkt_pts  = AV_NOPTS_VALUE;
+    pkt->dts  = av_rescale_q(ist->dts, AV_TIME_BASE_Q, ist->st->time_base);
 
     update_benchmark(NULL);
     ret = avcodec_decode_video2(ist->st->codec,
@@ -2623,16 +2667,15 @@ static int transcode_video(InputStream *ist, AVPacket *pkt, int *got_output, int
         /* no picture yet */
         if (!pkt->size)
             for (i = 0; i < ist->nb_filters; i++)
-                av_buffersrc_add_ref(ist->filters[i]->filter, NULL, 0);
+                av_buffersrc_add_ref(ist->filters[i]->filter, NULL, AV_BUFFERSRC_FLAG_NO_COPY);
         return ret;
     }
 
     best_effort_timestamp= av_frame_get_best_effort_timestamp(decoded_frame);
     if(best_effort_timestamp != AV_NOPTS_VALUE)
-        ist->next_pts = ist->pts = decoded_frame->pts = best_effort_timestamp;
+        ist->next_pts = ist->pts = av_rescale_q(decoded_frame->pts = best_effort_timestamp, ist->st->time_base, AV_TIME_BASE_Q);
 
     pkt->size = 0;
-
     pre_process_video_frame(ist, (AVPicture *)decoded_frame, &buffer_to_free);
 
     rate_emu_sleep(ist);
@@ -2732,7 +2775,6 @@ static int output_packet(InputStream *ist, const AVPacket *pkt)
 {
     int ret = 0, i;
     int got_output;
-    int64_t pkt_pts = AV_NOPTS_VALUE;
 
     AVPacket avpkt;
     if (!ist->saw_first_ts) {
@@ -2765,8 +2807,6 @@ static int output_packet(InputStream *ist, const AVPacket *pkt)
         if (ist->st->codec->codec_type != AVMEDIA_TYPE_VIDEO || !ist->decoding_needed)
             ist->next_pts = ist->pts = av_rescale_q(pkt->dts, ist->st->time_base, AV_TIME_BASE_Q);
     }
-    if(pkt->pts != AV_NOPTS_VALUE)
-        pkt_pts = av_rescale_q(pkt->pts, ist->st->time_base, AV_TIME_BASE_Q);
 
     // while we have more to decode or while the decoder did output something on EOF
     while (ist->decoding_needed && (avpkt.size > 0 || (!pkt && got_output))) {
@@ -2784,10 +2824,10 @@ static int output_packet(InputStream *ist, const AVPacket *pkt)
 
         switch (ist->st->codec->codec_type) {
         case AVMEDIA_TYPE_AUDIO:
-            ret = transcode_audio    (ist, &avpkt, &got_output);
+            ret = decode_audio    (ist, &avpkt, &got_output);
             break;
         case AVMEDIA_TYPE_VIDEO:
-            ret = transcode_video    (ist, &avpkt, &got_output, &pkt_pts);
+            ret = decode_video    (ist, &avpkt, &got_output);
             if (avpkt.duration) {
                 duration = av_rescale_q(avpkt.duration, ist->st->time_base, AV_TIME_BASE_Q);
             } else if(ist->st->codec->time_base.num != 0 && ist->st->codec->time_base.den != 0) {
@@ -2885,7 +2925,6 @@ static void print_sdp(void)
 
 static int init_input_stream(int ist_index, char *error, int error_len)
 {
-    int i;
     InputStream *ist = input_streams[ist_index];
 
     if (ist->decoding_needed) {
@@ -2912,17 +2951,6 @@ static int init_input_stream(int ist_index, char *error, int error_len)
         }
         assert_codec_experimental(ist->st->codec, 0);
         assert_avoptions(ist->opts);
-
-        if (ist->st->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
-            for (i = 0; i < nb_output_streams; i++) {
-                OutputStream *ost = output_streams[i];
-                if (ost->source_index == ist_index) {
-                    if (!ist->st->codec->channel_layout || !ost->st->codec->channel_layout)
-                        get_default_channel_layouts(ost, ist);
-                    break;
-                }
-            }
-        }
     }
 
     ist->next_pts = AV_NOPTS_VALUE;
@@ -3120,71 +3148,35 @@ static int transcode_init(void)
                 ist->decoding_needed = 1;
             ost->encoding_needed = 1;
 
-            switch (codec->codec_type) {
-            case AVMEDIA_TYPE_AUDIO:
-                ost->fifo = av_fifo_alloc(1024);
-                if (!ost->fifo) {
-                    return AVERROR(ENOMEM);
+            if (codec->codec_type == AVMEDIA_TYPE_VIDEO) {
+                if (ist && !ost->frame_rate.num)
+                    ost->frame_rate = ist->st->r_frame_rate.num ? ist->st->r_frame_rate : (AVRational){25, 1};
+                if (ost->enc && ost->enc->supported_framerates && !ost->force_fps) {
+                    int idx = av_find_nearest_q_idx(ost->frame_rate, ost->enc->supported_framerates);
+                    ost->frame_rate = ost->enc->supported_framerates[idx];
                 }
+            }
 
-                if (!codec->sample_rate)
-                    codec->sample_rate = icodec->sample_rate;
-                choose_sample_rate(ost->st, ost->enc);
-                codec->time_base = (AVRational){ 1, codec->sample_rate };
-
-                if (codec->sample_fmt == AV_SAMPLE_FMT_NONE)
-                    codec->sample_fmt = icodec->sample_fmt;
-                choose_sample_fmt(ost->st, ost->enc);
-
-                if (ost->audio_channels_mapped) {
-                    /* the requested output channel is set to the number of
-                     * -map_channel only if no -ac are specified */
-                    if (!codec->channels) {
-                        codec->channels       = ost->audio_channels_mapped;
-                        codec->channel_layout = av_get_default_channel_layout(codec->channels);
-                        if (!codec->channel_layout) {
-                            av_log(NULL, AV_LOG_FATAL, "Unable to find an appropriate channel layout for requested number of channel\n");
-                            exit_program(1);
-                        }
-                    }
-                    /* fill unused channel mapping with -1 (which means a muted
-                     * channel in case the number of output channels is bigger
-                     * than the number of mapped channel) */
-                    for (j = ost->audio_channels_mapped; j < FF_ARRAY_ELEMS(ost->audio_channels_map); j++)
-                        ost->audio_channels_map[j] = -1;
-                } else if (!codec->channels) {
-                    codec->channels = icodec->channels;
-                    codec->channel_layout = icodec->channel_layout;
-                }
-                if (av_get_channel_layout_nb_channels(codec->channel_layout) != codec->channels)
-                    codec->channel_layout = 0;
-
-
-//                 ost->audio_resample       = codec->sample_rate != icodec->sample_rate || audio_sync_method > 1;
-//                 ost->audio_resample      |=    codec->sample_fmt     != icodec->sample_fmt
-//                                             || codec->channel_layout != icodec->channel_layout;
-                icodec->request_channels  = codec-> channels;
-                ost->resample_sample_fmt  = icodec->sample_fmt;
-                ost->resample_sample_rate = icodec->sample_rate;
-                ost->resample_channels    = icodec->channels;
-                ost->resample_channel_layout = icodec->channel_layout;
-                break;
-            case AVMEDIA_TYPE_VIDEO:
-                if (!ost->filter) {
+            if (!ost->filter &&
+                (codec->codec_type == AVMEDIA_TYPE_VIDEO ||
+                 codec->codec_type == AVMEDIA_TYPE_AUDIO)) {
                     FilterGraph *fg;
                     fg = init_simple_filtergraph(ist, ost);
-                    if (configure_video_filters(fg)) {
+                    if (configure_simple_filtergraph(fg)) {
                         av_log(NULL, AV_LOG_FATAL, "Error opening filters!\n");
                         exit(1);
                     }
-                }
+            }
 
-                if (ist && !ost->frame_rate.num)
-                    ost->frame_rate = ist->st->r_frame_rate.num ? ist->st->r_frame_rate : (AVRational) { 25, 1 };
-                if (ost->enc && ost->enc->supported_framerates && !ost->force_fps) {
-                    int idx = av_find_nearest_q_idx(ost->frame_rate, ost->enc->supported_framerates);
-                    ost->frame_rate = ost->enc->supported_framerates[idx];
-                }
+            switch (codec->codec_type) {
+            case AVMEDIA_TYPE_AUDIO:
+                codec->sample_fmt     = ost->filter->filter->inputs[0]->format;
+                codec->sample_rate    = ost->filter->filter->inputs[0]->sample_rate;
+                codec->channel_layout = ost->filter->filter->inputs[0]->channel_layout;
+                codec->channels       = av_get_channel_layout_nb_channels(codec->channel_layout);
+                codec->time_base      = (AVRational){ 1, codec->sample_rate };
+                break;
+            case AVMEDIA_TYPE_VIDEO:
                 codec->time_base = (AVRational){ost->frame_rate.den, ost->frame_rate.num};
                 if (   av_q2d(codec->time_base) < 0.001 && video_sync_method != VSYNC_PASSTHROUGH
                    && (video_sync_method == VSYNC_CFR || (video_sync_method == VSYNC_AUTO && !(oc->oformat->flags & AVFMT_VARIABLE_FPS)))){
@@ -3234,7 +3226,7 @@ static int transcode_init(void)
                         size_t logbuffer_size;
                         if (cmdutils_read_file(logfilename, &logbuffer, &logbuffer_size) < 0) {
                             av_log(NULL, AV_LOG_FATAL, "Error reading log file '%s' for pass-2 encoding\n",
-                                logfilename);
+                                   logfilename);
                             exit_program(1);
                         }
                         codec->stats_in = logbuffer;
@@ -3383,15 +3375,6 @@ static int transcode_init(void)
                input_streams[ost->source_index]->st->index,
                ost->file_index,
                ost->index);
-        if (ost->audio_channels_mapped) {
-            av_log(NULL, AV_LOG_INFO, " [ch:");
-            for (j = 0; j < ost->audio_channels_mapped; j++)
-                if (ost->audio_channels_map[j] == -1)
-                    av_log(NULL, AV_LOG_INFO, " M");
-                else
-                    av_log(NULL, AV_LOG_INFO, " %d", ost->audio_channels_map[j]);
-            av_log(NULL, AV_LOG_INFO, "]");
-        }
         if (ost->sync_ist != input_streams[ost->source_index])
             av_log(NULL, AV_LOG_INFO, " [sync #%d:%d]",
                    ost->sync_ist->file_index,
@@ -3757,11 +3740,8 @@ static int transcode(void)
                     fclose(ost->logfile);
                     ost->logfile = NULL;
                 }
-                av_fifo_free(ost->fifo); /* works even if fifo is not
-                                             initialized but set to zero */
                 av_freep(&ost->st->codec->subtitle_header);
                 av_free(ost->forced_kf_pts);
-                swr_free(&ost->swr);
                 av_dict_free(&ost->opts);
             }
         }
@@ -4194,6 +4174,14 @@ static void add_input_streams(OptionsContext *o, AVFormatContext *ic)
 
             break;
         case AVMEDIA_TYPE_AUDIO:
+            guess_input_channel_layout(ist);
+
+            ist->resample_sample_fmt     = dec->sample_fmt;
+            ist->resample_sample_rate    = dec->sample_rate;
+            ist->resample_channels       = dec->channels;
+            ist->resample_channel_layout = dec->channel_layout;
+
+            break;
         case AVMEDIA_TYPE_DATA:
         case AVMEDIA_TYPE_SUBTITLE:
             if(!ist->dec)
@@ -4597,8 +4585,6 @@ static OutputStream *new_output_stream(OptionsContext *o, AVFormatContext *oc, e
         input_streams[source_index]->st->discard = AVDISCARD_NONE;
     }
 
-    ost->pix_fmts[0] = ost->pix_fmts[1] = PIX_FMT_NONE;
-
     return ost;
 }
 
@@ -4661,6 +4647,11 @@ static OutputStream *new_video_stream(OptionsContext *o, AVFormatContext *oc, in
 
         video_enc->bits_per_raw_sample = frame_bits_per_raw_sample;
         MATCH_PER_STREAM_OPT(frame_pix_fmts, str, frame_pix_fmt, oc, st);
+        if (frame_pix_fmt && *frame_pix_fmt == '+') {
+            ost->keep_pix_fmt = 1;
+            if (!*++frame_pix_fmt)
+                frame_pix_fmt = NULL;
+        }
         if (frame_pix_fmt && (video_enc->pix_fmt = av_get_pix_fmt(frame_pix_fmt)) == PIX_FMT_NONE) {
             av_log(NULL, AV_LOG_FATAL, "Unknown pixel format requested: %s.\n", frame_pix_fmt);
             exit_program(1);
@@ -4762,7 +4753,7 @@ static OutputStream *new_audio_stream(OptionsContext *o, AVFormatContext *oc, in
     audio_enc->codec_type = AVMEDIA_TYPE_AUDIO;
 
     if (!ost->stream_copy) {
-        char *sample_fmt = NULL;
+        char *sample_fmt = NULL, *filters = NULL;
 
         MATCH_PER_STREAM_OPT(audio_channels, i, audio_enc->channels, oc, st);
 
@@ -4775,22 +4766,23 @@ static OutputStream *new_audio_stream(OptionsContext *o, AVFormatContext *oc, in
 
         MATCH_PER_STREAM_OPT(audio_sample_rate, i, audio_enc->sample_rate, oc, st);
 
-        ost->rematrix_volume=1.0;
-        MATCH_PER_STREAM_OPT(rematrix_volume, f, ost->rematrix_volume, oc, st);
-    }
+        MATCH_PER_STREAM_OPT(filters, str, filters, oc, st);
+        if (filters)
+            ost->avfilter = av_strdup(filters);
 
-    /* check for channel mapping for this audio stream */
-    for (n = 0; n < o->nb_audio_channel_maps; n++) {
-        AudioChannelMap *map = &o->audio_channel_maps[n];
-        InputStream *ist = input_streams[ost->source_index];
-        if ((map->channel_idx == -1 || (ist->file_index == map->file_idx && ist->st->index == map->stream_idx)) &&
-            (map->ofile_idx   == -1 || ost->file_index == map->ofile_idx) &&
-            (map->ostream_idx == -1 || ost->st->index  == map->ostream_idx)) {
-            if (ost->audio_channels_mapped < FF_ARRAY_ELEMS(ost->audio_channels_map))
-                ost->audio_channels_map[ost->audio_channels_mapped++] = map->channel_idx;
-            else
-                av_log(NULL, AV_LOG_FATAL, "Max channel mapping for output %d.%d reached\n",
-                       ost->file_index, ost->st->index);
+        /* check for channel mapping for this audio stream */
+        for (n = 0; n < o->nb_audio_channel_maps; n++) {
+            AudioChannelMap *map = &o->audio_channel_maps[n];
+            InputStream *ist = input_streams[ost->source_index];
+            if ((map->channel_idx == -1 || (ist->file_index == map->file_idx && ist->st->index == map->stream_idx)) &&
+                (map->ofile_idx   == -1 || ost->file_index == map->ofile_idx) &&
+                (map->ostream_idx == -1 || ost->st->index  == map->ostream_idx)) {
+                if (ost->audio_channels_mapped < FF_ARRAY_ELEMS(ost->audio_channels_map))
+                    ost->audio_channels_map[ost->audio_channels_mapped++] = map->channel_idx;
+                else
+                    av_log(NULL, AV_LOG_FATAL, "Max channel mapping for output %d.%d reached\n",
+                           ost->file_index, ost->st->index);
+            }
         }
     }
 
@@ -4941,12 +4933,15 @@ static void init_output_filter(OutputFilter *ofilter, OptionsContext *o,
 {
     OutputStream *ost;
 
-    if (ofilter->out_tmp->filter_ctx->output_pads[ofilter->out_tmp->pad_idx].type != AVMEDIA_TYPE_VIDEO) {
-        av_log(NULL, AV_LOG_FATAL, "Only video filters are supported currently.\n");
+    switch (ofilter->out_tmp->filter_ctx->output_pads[ofilter->out_tmp->pad_idx].type) {
+    case AVMEDIA_TYPE_VIDEO: ost = new_video_stream(o, oc, -1); break;
+    case AVMEDIA_TYPE_AUDIO: ost = new_audio_stream(o, oc, -1); break;
+    default:
+        av_log(NULL, AV_LOG_FATAL, "Only video and audio filters are supported "
+               "currently.\n");
         exit_program(1);
     }
 
-    ost               = new_video_stream(o, oc, -1);
     ost->source_index = -1;
     ost->filter       = ofilter;
 
@@ -5127,7 +5122,7 @@ loop_end:
                 case AVMEDIA_TYPE_ATTACHMENT: ost = new_attachment_stream(o, oc, src_idx); break;
                 default:
                     av_log(NULL, AV_LOG_FATAL, "Cannot map stream #%d:%d - unsupported type.\n",
-                        map->file_index, map->stream_index);
+                           map->file_index, map->stream_index);
                     exit_program(1);
                 }
             }
@@ -5666,6 +5661,11 @@ static int opt_video_filters(OptionsContext *o, const char *opt, const char *arg
     return parse_option(o, "filter:v", arg, options);
 }
 
+static int opt_audio_filters(OptionsContext *o, const char *opt, const char *arg)
+{
+    return parse_option(o, "filter:a", arg, options);
+}
+
 static int opt_vsync(const char *opt, const char *arg)
 {
     if      (!av_strcasecmp(arg, "cfr"))         video_sync_method = VSYNC_CFR;
@@ -5848,8 +5848,8 @@ static const OptionDef options[] = {
     { "atag", HAS_ARG | OPT_EXPERT | OPT_AUDIO | OPT_FUNC2, {(void*)opt_old2new}, "force audio tag/fourcc", "fourcc/tag" },
     { "vol", OPT_INT | HAS_ARG | OPT_AUDIO, {(void*)&audio_volume}, "change audio volume (256=normal)" , "volume" }, //
     { "sample_fmt", HAS_ARG | OPT_EXPERT | OPT_AUDIO | OPT_SPEC | OPT_STRING, {.off = OFFSET(sample_fmts)}, "set sample format", "format" },
-    { "rmvol", HAS_ARG | OPT_AUDIO | OPT_FLOAT | OPT_SPEC, {.off = OFFSET(rematrix_volume)}, "rematrix volume (as factor)", "volume" },
     { "channel_layout", HAS_ARG | OPT_EXPERT | OPT_AUDIO | OPT_FUNC2, {(void*)opt_channel_layout}, "set channel layout", "layout" },
+    { "af", HAS_ARG | OPT_AUDIO | OPT_FUNC2, {(void*)opt_audio_filters}, "audio filters", "filter list" },
 
     /* subtitle options */
     { "sn", OPT_BOOL | OPT_SUBTITLE | OPT_OFFSET, {.off = OFFSET(subtitle_disable)}, "disable subtitle" },