Loudness

graywolf

graywolf

Well-known member
I have been eperimenting a bit with loudness, and have found out some interesting things.

I have been using Wavegain to remaster some CD's. Wavegain uses ReplayGain as it core. RG uses 89db (sound pressure) as its standard level. That is the equivalent of -14dbFS. Adding another -6db gives a neat -20dbFS loundness level which on my main Stereo plays nice with the volume know set to 9 o'clock. Now my immediate purpose is to have a CD Changer full of CD's that all play at the same level, but the following would seem to apply to our film sound levels as well.

Now we are talking about Loudness Levels here not Peak Levels. That is kind of like average levels, but not quite. I guess perceived levels would be the best way to describe it.

It would seem that just using the software to normalize the levels would be all that is needed. BUT, my experiments show that that is not so!

For instance a recording of Gershwin's Piano Rolls comes out too loud, I have to add -9db to it to keep it from blaring in relation to the other CD's. On the other hand, a recording of Doris Day with the band is too soft, I have to reduce the additional gain to -3db. Those examples are compared to the normal -6db additional gain for run of the mill recordings that sounds right.

The best I can figure is that the difference is caused by the sound density. The piano rolls are real dense, the levels change little, there is very little space between the peaks, and the lows are rather high. The Doris Day recording is the opposite, lots of level changes, lots of space between the peaks. You can see this in graphs of the sound (I usually use Audicity to check).

I guess what I am seeing here is that standards tell you one thing, but my ears (at least) tell something else.

I guess, what I have discovered here is that you almost never can get things right by simple measurements. On the other hand I have not come across anything so far that will not fit in that -3db, -6db, -9db (+/- 3db) set of adjustments.

ADDED: Just for information, I am adjusting the whole album (CD) not the individual tracks. Adjusting the album maintains the relationship between songs.
 
Last edited:
You keep talking about adding gain and then use -dB? At one point it sounds like you are boosting a level but you are going from -14 to -20 which would lower the level, by a lot. If you are adding gain they should be "+" not "-".

So I was confused by the specifics.

As a general thing you are doing what radio stations do. iTunes will also do this kind of thing. Both though would be doing it on a track by track basis.

VU is a better meter to be using for loudness. It is designed to more accurately reflect how humans hear. Most digital meters are more about monitoring the voltage/ bit levels rather than how loud it sounds. Probably the biggest factor in the differences are the age in which it was recorded. The piano scrolls though old were recorded recently after the age of loudness wars and Doris would have been recorded well before that.

And you are right that "loudness" has a lot to do with an average level so a varying dynamic will tend to sound lower than a less dynamic piece.
 
To measure loudness you should really use an appropriate instrument to measure it. For instance there is a loudness radar included in Adobe Audition or there are VST plugins like the ToneBoosters Loudness meter.
 
Scott, I use "adding gain" as used in Wavegain. But, adding a negative number is actually subtraction, if that confuses people. And, I guess, those piano rolls could be compressed to hell and back.

Cary, interestingly, I asked about this stuff over on gearslutz, and got nothing useful. All kinds of stuff on the web about normalizing peaks, not so much about normalizing loudness. What I am actually using was what I could find that did not cost me anything. Thanks for mentioning other posibilites.

For clairfication: What Wavegain does is adjust the loudness levels to an average of what you would hear in a movie theater.

Wavegain is a program that does the same as ReplayGain only to .wav files instead of the audio output. I get the impression it actually uses ReplayGain with a text interface, and then there is a wavegainfrontend.exe program that makes an easy to use semi-graphical interface. Kind of a Unix like philosophy to that.

The wiki on loudness states that 89db sound pressure level is the equivalent of -14dbFS. From that I substract 6db giving me -20dbFS, yes? Only wavegainfrontend.exe calls it additional gain, so adding -6db.

What Wavegain seems to do, as I am using it, is calculate the adjustment to each track, figure an average and apply that average to all the tracks in the album. You can also select to adjust each track individually if you wish, rather than averaging them.

I am to the point now where I can look at the waveform graphic of a couple of the tracks and pretty much guess what adjustment to use to get the same apparent level out of each CD. Since I have about 300 CD's it is a bit of a real project. I do know folks that have thousands of CD's. I would not even try to do that many.

I guess if you just want to playback mp3 rips, all you would need to do is use ReplayGain.
 
Steinberg's VST SLM128 loudness meter is a VST-3 plug, which is relatively new, so make sure your DAW supports it prior to investing time to downlaod install and tryin to figure out why it doesn't work. There's lots of other free meters available.
Most pro mastering engineers use their ears for matching song> song levels. Other than that, the loudness war goes on. A dynamic range meter is a good tool to have IMO.
 
No one has said it yet, so I will: ITU-R BS.1770-3. This is one of the main standards for trying to maintain loudness levels between different program material. As in, between a broadcast TV program and it's "interstitials" (commercials). What you need is a meter that reads in LUFS (loudness units full scale) which is also known as LKFS (loudness k-rated, full scale).

Here's a place to start researching.
 
You must log in or register to reply here.
Back
Top