On 2014-03-11 16:09:21 +0200, Martin Storsjö wrote:
> For live audio streams, requiring 500 frames for a stream to
> be detected is a bit overkill.
> 
> This allows live ADTS streams that aren't frame-aligned
> to start up more quickly, e.g.
> http://mp3.streampower.be/radio1.aac.
> ---
> Since the probe buffer size is increased until a good enough
> probe result is found, I'm not sure how much the previous
> patch actually helps though. If a false positive stream e.g. has
> got 200 matching frames but show mismatches after that, chances
> are that we will first try with a small buffer finding e.g. 70
> matching frames, which isn't enough. Next we double the probe
> buffer size and find 140 matching frames, which we deem enough -
> even if we would have realized that it's a mismatch if we would
> have tried with an ever bigger probe buffer, which we won't since
> we now regard 100 as enough.

I don't think that it will be a problem. What valid file would have
100 or more consectutive adts frames and then something else. I can
only think of diskimages containing adts files.

> ---
>  libavformat/aacdec.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/libavformat/aacdec.c b/libavformat/aacdec.c
> index 3aee82b..be213d4 100644
> --- a/libavformat/aacdec.c
> +++ b/libavformat/aacdec.c
> @@ -65,7 +65,7 @@ static int adts_aac_probe(AVProbeData *p)
>  
>      if (first_frames >= 3)
>          return AVPROBE_SCORE_EXTENSION + 1;
> -    else if (max_frames > 500)
> +    else if (max_frames > 100)

100 still seems very safe, maybe add a comment that it is just a random
number.

>          return AVPROBE_SCORE_EXTENSION;
>      else if (max_frames >= 3)
>          return AVPROBE_SCORE_EXTENSION / 2;
> -- 

ok

Janne
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to