Need definitive answer: exactly what does Loudness Normalization do in Audacity

HI all,
Due to ambiguities in various posts, audacity docs, and other articles, I need to get a definitive answer to a fairly simple (but multi-layered) question:

When applying Loudness Normalization in Audacity, is it only a linear adjustment of volume across the entire selected segment (in other words, is it strictly and only amplification)?

  • The use case is a music library of stereo tracks, using Audacity 3.x on Win10 64 bit
  • Are any changes made to EQ in the normalization process?
  • Is any compression applied?
  • Are any sneaky/clever algorithms applied to the content (that would modify psycho-acoustic affect of the music)

Thanks in advance for any answers/discussion on this topic.

It’s a simple linear adjustment. Amplify and the regular (peak) normalization are also linear.

The algorithm to calculate current (“before”) loudness is complex but it’s a straightforward adjustment.

However… It’s not checking the headroom or peak levels. If the loudness normalization pushes the peaks above 0dB and you export as a regular WAV file, you’ll get clipping (distortion) and that’s not linear. :wink:

Objectively there is no change to the EQ,
However, playing at a different volume will alter how the EQ sounds to the human ear: the louder the sound the flatter the ear’s EQ curve becomes … Equal-loudness contour - Wikipedia

Yeah, got it on clipping.
Good to know that the actual modification to the waveform is simple amplification, linearly applied.

Also, totally get it on the classic Fletcher-Munson curves. But that’s about “my perception of what the speakers produce” rather than “what has actually occurred to the input signal” – which, certainly appears to be amplification, consistently applied across the entire track.