Constant Gain Differential

Hi,
I’m looking for a way to continuously make the gain of Track B, 3 dB lower than Track A. In other words, I want the gain difference between them to be constant. How can I achieve this in Audacity?

Are you talking about the peak, the average, or perceived loudness?

The peaks are fairly easy… The Amplify effect will tell you how far you are from 0dB… If you run the Amplify effect and it defaults to +2dB, the current peak is -2dB, etc.

Once you know the peaks of the two files you can use Amplify to set the peak of the 2nd file 3dB below the peak of the first.

I’m looking for a way to continuously make the gain of Track B, 3 dB lower than Track A.

I read that as making track B pump in time to the volume on track A. I don’t know about forcing a 3dB change, tho. I think you can get AutoDuck to do something like that.

http://manual.audacityteam.org/man/auto_duck.html

I bet we’re too close. What’s the application? What are you doing? What’s the job?

Koz

Thank you both for the prompt reply.

DVDdoug, amplify I can do, and it’s the method I have been using so far. What I’m looking for now is some way to… constantly sample(?) both tracks, adjusting Track B to always be 3 dB below Track A. I don’t yet know enough about sound to find the solution on my own; but I know it can be done with hardware, so I imagine it’s possible with software.

Koz, the application is making subliminals for myself and my wife. Basically, I want this particular solution because I need the signal strength of the messages to be as loud as possible while still remaining undecipherable, and with amplify I can only go so loud before words or sentences are leaking through.

That’s going to be harder than you think. You want the loudness of the two presentations to maintain a relation to each other. Loudness isn’t one thing. It’s condition over time and there’s no shortage of different ways to calculate it and apply the results of the calculation.

For one example, we have a new tool for setting RMS (Loudness) for audiobook readings. Loudness is one of the specifications you have to meet for submission so someone will pay you. But it doesn’t change over time. It looks at your whole chapter, figures out the overall loudness, corrects it if needed, dusts off it’s hands and goes home. In general, that’s how the audiobok company does it, so everybody wins.

You can see if that helps you. Scroll down to where it says “Get and install RMS Normalize.” That’s the name of the tool.

https://forum.audacityteam.org/t/audiobook-mastering-version-4/45908/1

As a side note. You know most people can’t hear a 3dB change, right? That’s like changing the volume on your iPod 1/3 click.

Koz

Interesting… I did not know about loudness. But then I wonder, if RMS-Normalize can be calculated for a whole track, can a process like it instead be applied to every sample of that track?

Based on your explanation, it looks like the best solution for me would be to Limit the subliminal to keep it within a tight range, Limit the carrier track so it never fully drowns out the subliminal, and then apply RMS-Normalize to the sub track. What do you think?

Re 3dB: Just an arbitrary number to simplify my original question. I will find the right value by ear, but you are right to point that out; I imagine it will be closer to 20.

can a process like it instead be applied to every sample of that track?

Sure. Start the process now. Come back next week.

The reason it’s called “RMS” is the electrical association. Root Mean Square is an electrical measurement that calculates the area under the wave—the blue waves—like on the timeline. By definition, you have to have waves and waves are large collections of samples. I can’t think of a good stupid story but it’s like trying to derive the speed of a car by looking at the car very briefly in one place as it goes by. Not enough information. If you look multiple times, you can easily use that data to calculate the speed.

The reason we’re talking about this at all is that RMS turns out to be remarkably close to loudness in audio. It’s handy and they didn’t have to invent a whole new measurement.

So. What I would recommend missing any other ideas, is you create two carefully processed shows by using the AudioBook Mastering suite.

The top part of that instruction page is really all you need. All the rest of it is what to do if something goes wrong while you’re reading an inch-thick novel.

That should give you two carefully controlled, almost identical tracks. Then, use plain simple Effect > Amplify and reduce the background track by the amount you wish.

Select one track by clicking just above MUTE. Effect > Amplify: Amplification -20dB > Enter.

Koz