Tilt filter, "blue-ing" filter (cough cough martin?)

[guyman]

Hello all !

I was poking around and managed to get my hands on Martin's noise modules - pink, brown, and white. I noticed they were white noise oscillators, running through corresponding "pinking" and "browning filters" (-3 db, -6 db filters / octave respectively). I was wondering if we could collaborate to bring about "blueing" filters that could make blue noise (+3db) out of white, or white out of pink or brown noise... or perhaps a freely adjustable perfect tilt filter in general. I had reached to Spogg, but the wise man said I should share this idea publicly, and that our great savior Martin would be the gentleman who would more than likely bring insight into the matter. I'm not a coding genius with filters, so hopefully we can all learn and benefit from this creation.

[martinvicanek]

I can offer these . I might do a general purpose tilt filter some time where you can vary the slope within, say, +- 12 dB/octave with a sigle knob...

[guyman]

Wow thanks!!

A general tilt filter would be boss... what's the slope on these different curves?

[martinvicanek]

Here ya go - spectral tilt filter. The transfer curve is an approximate power law over the entire audio spectrum, with powers selectable within the range -3/2 to +3/2, or, equivalently, +-9 dB per octave (or +-30 dB per decade). You can stretch the range a bit but those would be extreme settings really.

The filter is a 3pole/3zero IIR , where the poles and zeros are suitably dragged aroud as you turn the knob. Normalization is chosen such that if you feed in pink noise, you get constant output power whatever the tilt.

The device would work in poly or in stream, or packed stream for stereo.

Have fun!

P.s. Uploaded improved version 2:
- extended range -15 dB/octave to +15 dB/octave
- more accurate power law approximation
- improved ultra-low frequency response
- optimized filter
- suppressed denormals

[RJHollins]

Very nice !

Thanks Martin V.

[adamszabo]

Thats really nice Martin! Is it possible to adjust the frequency of the middle point where the tilt occurs?

[guyman]

Martin. You are in full beast mode at all times. Thank you !!

[guyman]

You could do that adam, but I believe it would defeat the purpose of the perfect tilt per decade which is consistent across the spectrum... needed to perfectly "pink", "brown", "blue", or "white" a noise profile.. (or signal )

[guyman]

This is my (failed) attempt at using this exact tilt to create a "simultaneous tilt filter" that would attain the perfect "smile" or "frown" shape to apply to the signal which would "brown/pink" the highs while "blueing/whiting" the lows - and vice versa. I believe phase and other things I'll attribute to my lack of cleverness is defeating this idea before I can get it out the gate....

[guyman]

Seeing as how we use these 4 noise profiles (white,pink,brown, blue) to tune/test systems and signal chains... creating a multipurpose tool that allows us to manipulate the spectrum consistent to this "perfect" tilt, satisfies the portion of my mind that craves a consistent control/metric at which I like to quickly alter the spectrum. I need to sit down and study filters so I can be more like Martin...