[FFmpeg-devel] [PATCH 2/3] lavfi/loudnorm: add an internal libebur128 library

[Marton Balint]

On Thu, 3 Nov 2016, Kyle Swanson wrote:

> Hi,
>
>> On Sun, Oct 16, 2016 at 2:32 PM, Marton Balint <cus at passwd.hu> wrote:
>
> I haven't been checking ffmpeg-devel too much lately and just saw
> these patches today. I'm glad to see that an ffmpeg libebur128 port is
> being taken seriously now, I had some trouble getting traction on a
> few similar patches earlier this year. A new version of libebur128 was
> released a few days ago, so we can probably improve this port even
> more. The loudnorm filter continues to be a popular ffmpeg feature. I
> still get messages about af_loudnorm from users all the time, so it'd
> be nice to remove the external library dependancy and open this up to
> everyone.
>> Also contains the following changes to the library:
>> - add ff_ prefix to functions
>> - remove cplusplus defines.
>> - add FF_ prefix to contants and some structs
>> - remove true peak calculation feature, since it uses its own resampler, and
>>   af_audnorm does not need it.
> I'd argue that we should include the true peak calculation because
> after this port we'll want to replace the code in f_ebur128 as well.
> Libebur128 v1.2 drops the speex resampler and includes its own small
> resampler (fast and passes all the EBUR128 tests.) Also there's a
> patch I sent in June which uses libavresample APIs.

Yeah, I started the porting work on a version which already had it's own 
resampler, however I decided to strip down as many things as possible 
(which we not use), to make life easier. Also current version of the 
library would force you to use loudness measurement even if you only want 
true peak, which is not good performance-wise, so it needs some additional 
work to be implemented properly. Anyway, it can be added later to the 
codebase (if something needs it).

Regards,
Marton