musical thoughts

More reposted material!

If you haven’t already, please have a look at dimensions of tone in which I explain the background of what is going on here.

In part three, we’ll try to measure the precise effect of the tone controls on a suitcase (sorry stage owners–it’s what I have on hand).

To accomplish this, I recorded a test signal: a .wav file with 10 seconds of pink noise. Pink noise is widely used as a test signal since it has 1/f power distribution. You can translate this into a useful musical term–every octave contains the same amount of energy. I played the test file out of my audio interface and into the effect loop return of the suitcase preamp (and hence through the amp and modified speakers). I measured the response with a Behringer ECM8000 reference microphone, which has a reasonably flat response curve (you can look up the specs if you like, but it is “flat enough” within 20Hz-20kHz). That signal was sampled back in my Edirol UA-25 at 48kHz and 24 bit sample size (as previous measurements). The graphs were produced with Audacity.

First let’s examine the spectrum of the input signal. On a log-log scale, you’d expect 1/f to be a straight line.

Looking good. That’s a 3dB/octave slope. Now let’s have a peek at the output spectrum with the tone controls set flat:

I overlaid this and the others on the input curve for easy comparison. What can we observe?

• bass response falls off rapidly below 45Hz. Lowest note on a seventy-three is about 41 Hz.
• From say 100Hz to 3kHz, the response tracks 1/f pretty well overall.
• There are a couple [b]very[/b] prominent notches in the spectrum at 359Hz (F4) and 1423Hz (F6)
• Above 5kHz things roll off quite a bit more steeply than 3dB/octave. Yours will roll off even faster since I modified mine.

So the suitcase bottom colors the sound substantially. We knew that already, but this gives us a way to visualize it. The bass is boosted somewhat even though the tone controls are set flat.

Next I pegged the bass tone control. I was a little surprised at the resulting response:
]

You get a 6dB hump centered at around 91 Hz that’s a couple octaves wide. What surprised me was the apparent cut in treble that came along for the ride. You wouldn’t expect that from a graphic equalizer–I never considered that the bass boost may also cut treble.

Next I pegged the treble.

This seemed to take effect in the octave between 3kHz and 6kHz entirely. The EQ may actually have a wider band, but the rolloff of the speakers is swamping any effect that would have. And look at that, a corresponding bass cut (look at 95Hz–we lost 6db compared to the flat curve). Very interesting.

And then I pegged both bass and treble:

Interestingly, but not surprisingly now that we’ve seen each control separately, this results in something a lot like the flat curve, but with the middle dropped out a bit. I don’t view this as a practical setting, I just threw it in for interest.

I’m working on getting you a good sound file to listen to the coloration (music, not pink noise). Bear with me while I work out mic placement issues.

Let me know if this kind of article is useful. (Originally posted on the SuperSite Tue Jul 31, 2007)