This is a very simple patch that DISTORTS your guitar (or any other instrument, or input, that you want), but it does so in the following way:

• if you play pp (pianissimo), the sound will be pp and distorted.
• if you play ff (fortissimo), the sound will be ff and distorted.

Let's explain it differently: you'll have distortion but at the same level as your input, i.e. NOT louder than what you played, as is usually the case.

As you can imagine, this allows for expressive dynamic playing (or processing).

Download:

• Warning: if you don't have much experience with Pd, spend a couple of minutes and make sure your system or headphones cannot output loud sounds.

Expressive Distortion.pd

Extra info:

The origin of this patch is in the following thread: http://forum.pdpatchrepo.info/topic/9073/looking-for-a-distortion-with-same-level-as-input/5

More extra info:

I have uploaded a similar patch here:
http://forum.pdpatchrepo.info/topic/9088/expressive-distortion-by-transfer-function

Bug? Comments? Post them here. Or share your results at http://forum.pdpatchrepo.info/category/3/output

Cordially, Landon

@LandonPD I am curious -- what is the significance of multiplying by the square root of two?

Nice question!

There's a little explanation in the [pd FAQ] object, which you've probably already read.

The last sentence should be replaced by something else, e.g.:

[square root of 2]······>amplitude compensation factor (so that a full blast sinusoid input, with a distortion amount of 1 [no distortion], is output as a ±1 signal and not as a ±0.707 signal).

INPUT: SINUSOID (distortion amount: 1, this means no distortion)

When you input a full blast sinusoidal test signal, you get the following readings:

• [env~] ≈ 97
• [dbtorms] ≈ 0.707
• [* 1.4142] ≈ 1

INPUT: SQUARE WAVE (distortion amount: 1, this means no distortion)

That's all great...our input was ±1 and our scaling factor also became 1. But this is only valid for a sinusoid, not for other waveforms. When you input a full blast square wave test signal, you get the following readings:

• [env~] ≈ 100
• [dbtorms] ≈ 1
• [* 1.4142] ≈ 1.4142

Is this 1.4142 scaling factor too big for our needs? Hm...

Our original square wave input becomes louder than it originally was. But this patch assumes that you are playing guitar, and that your guitar's signal will never be a full blast digital square wave (what kind of crazy guitar would that be?). So the 1.4142 scaling factor is right, in general.

So I imagine this satisfies your curiosity, multiplying by the square root of two is for amplitude compensation, for boosting the signal.

One could delete [* 1.4142] and forget completely about it. That's would be useful too. However, deleting [* 1.4142], and using instead the scaling factor produced by [dbtorms], would only produce a scaling factor of 0.707 (when your input is a full blast sinusoid)...0.707 is too small...I want a factor that is larger than 0.707. So my solution is multiplying 0.707 by 1.4142. In this way, (when your input is a full blast sinusoid) the scaling factor becomes 1, which is more useful than the weaker 0.707.

Another extra issue is the final clipping function, [clip~ -0.707 0.707]. I guess one should delete it, or set it to [clip~ -1 1]. Hm...