[FFmpeg-devel] [PATCH] avformat/mp3: skip junk at the beginning of mp3 files

wm4 nfxjfg at googlemail.com
Wed May 27 14:48:10 CEST 2015

Apparently it can happen that a mp3 file has junk data between id3 tag
and actual mp3 data. Skip this to avoid outputting nonsense timestamps.
(Two packets had the same timestamps, because the mp3 parser failed to
compute a frame duration.)

In this case, the junk consisted of 1044 bytes of zero, which
incidentally is the same size as normal mp3 frames in this stream. I
suspect the mp3 was edited with some tool which wiped the Xing/LAME
headers. Data near the end of the file suggests it was encoded with
"LAME3.97", but the normal Xing/LAME headers are missing. So this could
be "normal". mpg123 also attempts to skip at least 64KB of junk data by
scanning for headers.
This assumes the stream buffer is not smaller than 64KB + max. mp3 frame
size, or the seeks could turn into a problem.
 libavformat/mp3dec.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/libavformat/mp3dec.c b/libavformat/mp3dec.c
index abd7fc0..a5374a1 100644
--- a/libavformat/mp3dec.c
+++ b/libavformat/mp3dec.c
@@ -54,6 +54,8 @@ typedef struct {
     int is_cbr;
 } MP3DecContext;
+static int check(AVFormatContext *s, int64_t pos);
 /* mp3 read */
 static int mp3_read_probe(AVProbeData *p)
@@ -370,6 +372,16 @@ static int mp3_read_header(AVFormatContext *s)
     if (ret < 0)
         return ret;
+    off = avio_tell(s->pb);
+    for (i = 0; i < 64 * 1024; i++) {
+        if (check(s, off + i) >= 0) {
+            av_log(s, AV_LOG_INFO, "Skipping %d bytes of junk at %lld.\n", i, (long long)off);
+            avio_seek(s->pb, off + i, SEEK_SET);
+            break;
+        }
+        avio_seek(s->pb, off, SEEK_SET);
+    }
     // the seek index is relative to the end of the xing vbr headers
     for (i = 0; i < st->nb_index_entries; i++)
         st->index_entries[i].pos += avio_tell(s->pb);

More information about the ffmpeg-devel mailing list