[FFmpeg-devel] [PATCH] avfilter: add loudnorm
h.leppkes at gmail.com
Sat Apr 9 01:06:41 CEST 2016
On Fri, Apr 8, 2016 at 11:15 PM, Kyle Swanson <k at ylo.ph> wrote:
> On Fri, Apr 8, 2016 at 4:07 PM, Kyle Swanson <k at ylo.ph> wrote:
>> On Wed, Apr 6, 2016 at 8:09 PM, Carl Eugen Hoyos <cehoyos at ag.or.at> wrote:
>>> Kyle Swanson <k <at> ylo.ph> writes:
>>>> Does anyone have any comments about this besides the fact
>>>> that it links an external library?
>>> It is not ok to add an additional dependency to FFmpeg if the
>>> features of the new dependency are already available within
>>> (Note that I cannot judge if this is the case for your patch.)
>> Any issues with breaking out the measurement logic from the ebur128
>> filter into utility functions? I can take a crack at it if no one
>> thinks it's a horrible idea...
> Or adapt the libebur128 code, which has a couple of nice features and
> an MIT license. Obviously the oversampling would have to be rewritten
> to use swresample.
In case you didn't pick up on that particular part, but having the
same basic functionality twice in different code is what the entire
discussion here is trying to avoid.
So if you can reuse the existing ebur128 code that Clement wrote for
the analysis filter that would be the best situation.
Apart from that, the filter looks extremely long for doing a lot of
the same things the r128 filter we have already does, couldn't it just
be build on top of the analysis the existing r128 filter already
provides, and just adjust the audio based on its data?
More information about the ffmpeg-devel