[FFmpeg-user] FFMPEG loudnorm - linear normalization

CMG DiGiTaL cmarciog at gmail.com
Wed Mar 9 13:14:36 EET 2022

> Linear processing is multiplying whole file with single float number,
> volume/gain adjustment.
> Dynamic processing is changing dynamics of input audio.

hi Paul,

I understand what you said, but as I mentioned, I have some doubts related
to the test I did on the WLM Plus Stereo

When I run the plugin it first analyzes all the audio and has a GAIN/Trim
function, which at the end of the analysis
shows a value to be corrected so that the audio reaches the LUFS that was

The value generated by the GAIN/Trim function in the audio that in my batch
does not
 reach the target value of -10 LUFS, is *+3.8*.

Enabling this gain (+3.8), the plugin manages to make the audio reach the
-10 LUFS that I set.

Audio information I run in the plugin, putting the target LUFS -10.0 and
the True peak -0.5:
   Input Integrated:    *-17.2* LUFS
   Input True Peak:      *-1.1* dBTP
   Input LRA:             *3.6* LU
   Input Threshold:     *-27.6* LUFS

   Output Integrated:   -13.8 LUFS
   Output True Peak:     -0.5 dBTP
   Output LRA:            3.8 LU
   Output Threshold:    -23.8 LUFS

   Target Offset:        *-0.4* LU

Note: In my batch, with these values above, the final result is -13.5 LUFS.

I have two questions:
1 - Analyzing the audio values above, we can see that the gain that the
plugin applies is the same value as the
    Output LRA, does this mean that the generated gain is applied to the
LRA value?... if I want my batch to also
    reach the - 10 LUFS I have to put the LRA as +7.6, is that it?

2 - Is the plugin doing this normalizing in a linear or dynamic way?... is
it compromising the audio to achieve what
     I defined as a target?

More information about the ffmpeg-user mailing list