Mixing Test Tones


While trying to balance my orchestral template I bumped into something odd which I don’t understand.

Start with an empty project. Add a stereo audio track. Insert the included Test Tone Generator with the default settings (sine wave, -12db). Duplicate the track three times, for a total of four tracks playing the same test tone together.

The level I get at the output varies widely when I mute some of the tracks. Here’s what I get when each track is active (i.e. unmuted):

1       -12
1+2     -11,2
1+2+3   -7,8
1+2+3+4 -4,8

2+3+4   -5,2
3+4     -10
4       -12

1+4     -6,3
1+3+4   -7,5
1+2+4   -5,8

Also, note that when going from 1+4 to 1+2+4 I actually get a lower output level.

Could someone please explain what is happening here?


Assumption: Test tones are not exactly the same frequency, and/or the 4 tracks’ sine waves aren’t synchronized (and therefor cancel each other out in a partial or less partial way). Phase is the magic word here. If you’re interested, you should try the same experiment with white or pink noise. Also - always mention what type of reading you use at the stereo output meters. ‘Level’ and then dBs is not enough to describe it.

The sine test tones are the same frequency (the default one, 440 Hz). But you are right about the phase issue, because apparently the oscillators are not in sync. I tried the same test with white noise and got consistent results (-12, -6, -2,5, -0,2 for 1, 2, 3 and 4 channels respectively).

The reading I was referring to is the output meters’ Peak Meter Value (which I was resetting after each test).

What I’m trying to figure out is this: if I have four channels going into a bus all at the same fader value, and then I reduce each of the four faders by, say, 2db. Now, by resetting the four faders to their original value, by how much would I need to attenuate the bus fader in order to get the same result? What is formula for this?


Interesting to know that the test tones each live a life of their own… Before this arose, I’d assume they would all start at the same zero point and stayed in sync just from the processor clock - well, apparently not. As for the formula, it’s not rocket science, but I have to dig too deep to give you an answer off-hand :confused: