Does Normalization Boost EQ?

I’ve notice that previewing before normalization in Audacity, the EQ seems to boost? The same experience with Mp3DirectCut. I have made sure that when testing, the audio output levels are closely same from an non-normailzed and normalized MP3 file. I have read that normalization only affects the ouput levels and not EQ or dynamic range but I’ve also read where normalization can boost the EQ as well. The samples were created from normalizing Amazon MP3 songs.

This topic has less to with the OS and more on the impact of normalizing. Perhaps it could be moved to a more appropriate forum.

Normalizing in Audacity (and most other audio programs) affects only the amplitude of the audio.
I can’t speak for all other audio programs.

Than what could account for the boost in EQ? I have to trim the bass down to get a more flat EQ, where this is needed on the pre-normalized version of the song. Could it be any setting or a use of a different CODEC?

Thank you for your response.

It is a psychoacoustic phenomenon.
Not all frequencies appear with the same loudness for different volume settings.
There are the so called equal loudness curves, where all frequencies are apparently equal loud in comparison to the 1 kHz tone.
However, this applies only when the output gain is 70 dB or so.
Hence, you have always to make compromises in your mix because you don’t know on which devices and with what volume the final track will be played.

Thank you for your explanation. This seems to cancel the benefits of most normalization processes.

This seems to cancel the benefits of most normalization processes.

Only if you insist on the literal meaning of “normalize.” In the electronic context, it just means make the blue waves bigger until you run out of room. Performance loudness is irrelevant. It’s a very close cousin to Amplify. They have slightly different tools and there was discussion about making them one. It would not be that hard.

You’re probably think more along the lines of Chris’s Compressor. He designed an Audacity plug-in version of Broadcast Compression which smashes everything to one volume level including volume variations within a song. There is a compression slider which gives you control of how vigorously it does this.

And yes, you ran into the Salesroom Effect. Make the music system you want to sell the loudest one.

Koz

Thank you Koz,

My application is geared for portable mpg3 players like the Sandisk Sensa Clip Disk. For those music files which have low levels, the device spends more battery life on having to raise to volume levels, also some songs are high volume while others are much lower.

There is another issue when listening to classical music where many times passages are low while other are thundering. How does normalizing impact this type of situation? I will look at Chris’s Compressor and experiment with that product.

Loudness and volume are 2 different phenomenons. Loudness is not factored into normalizing, like you said - it is just raising the blue graph.

Chris designed his compressor so he could listen to opera in the car. He said so. It is as successful as it is because he started with a musical performance and a goal and worked backwards rather than a whiteboard full of algorithms and working forward. I use it once a week to make a talk show useful in the field. It has two performers, one has a nuclear laugh and the other mumbles in his beer. I advance the first slider, Compression from 0.5 to 0.77.

These are the three wave samples: No compensation, Default 0.5 and my 0.77 settings.

http://kozco.com/tech/audacity/pix/ChrisCompressorBeforeandTwoAfters.png

MP3’s family name is MPEG-1, Layer 3. It’s the sound compression for a very old video standard.

I never met a Personal Music Player which would go flat faster with quiet music. But yes, if I simply drag podcasts onto my iPod, the volume variations within a show are pretty serious.

“OK, now I’m going to wait until I get to a quiet street to play that last segment again and hear what they said.” I’ve been known to intentionally walk out of my way to avoid noisy roads.

Koz

Normalization in Audacity is mainly for equalizing the left and right channels of a stereo track to the same level. Amplifying one way or the other is more of a byproduct and is optional. I’ve never monitored it, but in my experience, raising or lowering the amplitude significantly does seem to change the perceived equalization.

The Normalize effect is described in detail in the manual here: http://manual.audacityteam.org/o/man/normalize.html

Lots of information about “loudness” here: http://en.wikipedia.org/wiki/Loudness
Particularly relevant to this conversation is the “Equal loudness contour” http://en.wikipedia.org/wiki/Equal-loudness_contours

I’ve never monitored it, but in my experience, raising or lowering the amplitude significantly does seem to change the perceived equalization.

Possibly less about “Perceived EQ,” (loudness) and with more on the actual change of the EQ bias. The volume levels for each, before and after normalization changed the EQ.

About Chris’s Compensator:

Theoretically the way this works it seems that you would loose the dynamics of the soft and loud passages experienced in classical music. I did used it and it seemed not to change the dynamics. Will test more.

As you can see here, the Normalization effect does not change the Eq in any way: Google Code Archive - Long-term storage for Google Code Project Hosting.

Just in case it’s not clear… The loudness curves are an ACOUSTIC effect that depends on the LOUDNESS of the sound hitting your EARS.

If you turn-up the volume, the bass SEEMS to be turned-up more. If you turn-down the volume, the bass SEEMS to be turned-down more.

This has NOTHING to do with audacity or the volume in the digital file. If you reduce the gain by 6dB in Audacity and then turn-up the playback volume by 6dB, there will be no change in loudness and no perceived change in EQ/frequency balance.

Possibly less about “Perceived EQ,” (loudness) and with more on the actual change of the EQ bias. The volume levels for each, before and after normalization changed the EQ.

There is no change in EQ. Period. End of story.

It’s not easy, in fact IT’S IMPOSSIBLE to accidently change the EQ of a digital file.

About Chris’s Compensator:

Theoretically the way this works it seems that you would loose the dynamics of the soft and loud passages experienced in classical music. I did used it and it seemed not to change the dynamics. Will test more.

Did you try opera or classical music? If there are no dynamics to begin with it’s harder to boost the dynamics.

Or, try experimenting with the ratio or hardness controls.