Normalize question

I’ve been using Amplify to increase the volume of my cassette-to-cd conversion project and verified that Amplify analyzed the tracks in my project (ie. 5 tracks/songs from 1 side of a cassette) and increased them all but at different db’s. Example: If I set Amplify to 0 I will find at least 1 track/song was increased to 0 but others are less which I assume Amplify would do, boost the loudest part or track to 0 and other, quieter songs will be less than 0. Again, makes sense.
So I just tried Normalize, set the max amp to 0, and checked the Normalize channels independently and expected all tracks to increase to 0 but they do not, they are different still with at least 1 bossting to 0. I thought this setup would act like Media Players “leveling across tracks” does (or I thought it did), boosting all tracks to the same db. Not that I want to make all my tracks the same volume since they were not originally recorded as such. But I was testing and thought Normalize would make all the tracks the same volume that I chose…

They’re both arithmetic tools. They ignore the music and make the highest peak in the work you select come out to 0 if you left the defaults. Amplify does it the same to the whole show and Normalize does it left and right independently.

That’s pretty much it. This kills people trying to transfer vinyl because any cat hair pop on the record can kill both tools.

The tools are more complex than that because you can cause Amplify to intentionally create overload damage and I believe you can cause Normalize to link left and right now. Normalize also has the DC Removal tools. None of that changes what they’re thinking about. Neither one has any idea of the musical content of the show.

If you want to make lower volume sounds louder at the expense of higher volume sounds in the same show, then neither of these is for you. You want one of the compressors or other dynamic volume managers. Those tools do make the effort to know what the music is and change the show so the music is actually different than when it started.


Hey PCat, I had a similar problem, check out answers given, to my post of Oct 10:

"Avoiding compression, but raising level "

Here’s a link to the topic that wborys is referring to:

The “standard conservative” approach is to combine both sides of the cassette/LP into one big file. That way, any volume changes are applied equally to all songs so the relative volumes don’t change… Loud songs remain loud, quiet songs remain quiet, etc.

With an analog original, there may be only one song with one 0dB peak after normalizing the album as a whole. There are a couple of reasons for this - There is not as much precision in analog… There is no absolute hard-limit in analog… The analog recording/playback process tends to boost some peaks and attenuate others (MP3 encoding also does this). Most modern music is (dynamically) compressed* and peak-limited to make it as loud as possible. So on a modern CD, you may find that all (or most) tracks have peaks of exactly 0dB. The typical exception would be a jazz or classical CD with less digital processing and lots of remaining dynamic contrast (loud and quiet parts).

As Koz says, the peak level has little correlation with perceived loudness. A quiet-sounding song can have higher peaks than a loud-sounding song.

Loudness is more related to the average level and the frequency content. BTW - If you increase the peak by some amount, say 3dB (linearly without compression), the average will also increase by 3dB.

Dynamic compression can be used to boost the average level without boosting/distorting the peaks. But of course, this reduces the dynamics of the music and it can become boring.

There is a tool called ReplayGain (and it’s variations) that tries to match percieved loudness of your songs. Because many of your quiet-sounding songs can’t be boosted without clipping/distortion, ReplayGain can only match the volumes by making loud songs quieter. That frustrates some users who really want to make their quiet songs louder.


  • Dynamic compression is completely different from file compression (such as MP3).