avconv: Handle audio sync for non-S16 sample formats.

Also fix reporting of the number of samples added/dropped.

avconv.c

@@ -729,6 +729,14 @@ static void write_frame(AVFormatContext *s, AVPacket *pkt, AVCodecContext *avctx
     }
 }
 
+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 void do_audio_out(AVFormatContext *s,
                          OutputStream *ost,
                          InputStream *ist,
@@ -829,9 +837,9 @@ need_realloc:
 
     if(audio_sync_method){
         double delta = get_sync_ipts(ost) * enc->sample_rate - ost->sync_opts
-                - av_fifo_size(ost->fifo)/(enc->channels * 2);
-        double idelta= delta*dec->sample_rate / enc->sample_rate;
-        int byte_delta= ((int)idelta)*2*dec->channels;
+                - 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){
@@ -840,7 +848,8 @@ need_realloc:
                     byte_delta= FFMAX(byte_delta, -size);
                     size += byte_delta;
                     buf  -= byte_delta;
-                    av_log(NULL, AV_LOG_VERBOSE, "discarding %d audio samples\n", (int)-delta);
+                    av_log(NULL, AV_LOG_VERBOSE, "discarding %d audio samples\n",
+                           -byte_delta / (isize * dec->channels));
                     if(!size)
                         return;
                     ist->is_start=0;
@@ -854,11 +863,11 @@ need_realloc:
                     }
                     ist->is_start=0;
 
-                    memset(input_tmp, 0, byte_delta);
+                    generate_silence(input_tmp, dec->sample_fmt, byte_delta);
                     memcpy(input_tmp + byte_delta, buf, size);
                     buf= input_tmp;
                     size += byte_delta;
-                    av_log(NULL, AV_LOG_VERBOSE, "adding %d audio samples of silence\n", (int)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);
@@ -871,7 +880,7 @@ need_realloc:
         }
     }else
         ost->sync_opts= lrintf(get_sync_ipts(ost) * enc->sample_rate)
-                        - av_fifo_size(ost->fifo)/(enc->channels * 2); //FIXME wrong
+                        - av_fifo_size(ost->fifo)/(enc->channels * osize); //FIXME wrong
 
     if (ost->audio_resample) {
         buftmp = audio_buf;
@@ -1436,14 +1445,6 @@ static void print_report(OutputFile *output_files,
     }
 }
 
-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 void flush_encoders(OutputStream *ost_table, int nb_ostreams)
 {
     int i, ret;