[FFmpeg-user] LUFS Normalization of WAV files

CMG DiGiTaL cmarciog at gmail.com
Thu Mar 31 01:39:18 EEST 2022


Em qua., 30 de mar. de 2022 às 07:27, CMG DiGiTaL <cmarciog at gmail.com>
escreveu:

> Better use ebur128 scanner filter, it is much faster, after it reports
>> values, just apply volume filter.
>>
>
>
>> loudnorm dynamic processing is not perfect.
>
>

> hi Paul,
> Really?... based on the command line I sent, how could I do it the way you
> are saying?... can I use the same
> command changing some parameters or would it be a completely different
> command? How can I do it the way
> you suggested?
>
> thanks
>

hi Paul,

I've already managed to create a bat using ebur128, ok?.

I did some tests too and didn't see a big difference with loudnorm, but
I'll keep testing other audio formats with it.
 Below is one of the tests I did:

ffmpeg -hide_banner -i "C:\Users\CMG\Desktop\WAV audio LUFS\audio.wav" -af
ebur128 -f null NUL

Integrated loudness:
I:          *-8.8* LUFS
Threshold: *-18.8* LUFS

Loudness range:
  LRA:         2.0 LU
  Threshold: -28.8 LUFS
  LRA low:    -9.7 LUFS
  LRA high:   -7.8 LUFS

-----

ffmpeg -hide_banner -i "C:\Users\CMG\Desktop\WAV audio LUFS\audio.wav" -af
"[0:a]loudnorm=print_format=summary" -f null NUL

Input Integrated:     *-8.9* LUFS
Input True Peak:      +0.7 dBTP
Input LRA:             2.0 LU
Input Threshold:     *-18.9* LUFS

Output Integrated:   -23.8 LUFS
Output True Peak:    -13.8 dBTP
Output LRA:            1.7 LU
Output Threshold:    -33.8 LUFS

Normalization Type:   Dynamic
Target Offset:        -0.2 L

if we analyze the Threshold inputs, the difference is 0.1dB of loudness. I
don't know if in an example with another song, it might actually
make a bigger difference, but in this case, there wasn't a significant
difference!

thanks


More information about the ffmpeg-user mailing list