What would be the most simple way of getting a patch to start playing white noise, and then over a pre-determined length of time, for it to filter into a sine wave?

I'm thinking that using the [bp~] object to filter the [noise~] through to a very narrow "Q" range may do this... or an approximation at least.

an ever narrowing band-pass filter. but i doubt you'll get a perfect sine tone out of it. also i guess there would be technical time dependant issues to do with this, so you may want to also mix in a sine oscillator tuned to the same frequency as the filter

this is sort of what you describe. Does it make sounds you imagine?

http://www.pdpatchrepo.info/hurleur/nois2sin.pd

Kind of... it's a good basis for me to play around with, so thanks.

It seems from this and from my own attempts that the "sine" using the Q range is never very pure as there are always some modulation artifacts left from the [noise~] no matter how tight the Q range is.

It could be that a more algorythmic program like SuperCollider is better suited to this sort of thing.

if the result is still to noisy you could use 3 or 4 bp~ in serial to make the filters slew rate more steep.

you can also fake it a little with an [osc~] that you bring up while filtering in.

I don't think its realistic to filter digitally created white noise and produce a beautiful sine tone. The theory is that there will be a standard distribution of energy across all frequencies, but in practice the distribution is not constant over time. just my opinion on digital noise.

Maybe the way to approach this is to do it all in the spectral domain, ie create the 'noise' as a constant of say 512 bins and then slowly filter the bins to one. There might actually be one of the FFTease objects already for this.....

This is of course all off the top of my head!

The spectral domain idea may be well worth a look - what I have found is that using the chebyshev filter in the audio examples folder, you can go form a sort of "full bandwith" noise effect down to a pure sinewave fairly easily.

Bit cheaty, but it's the closest to my idea at the moment.

i made a patch that had 2 tables, and used fft to get between them. The problem here is that the sine wave will be in one band, while the noise occupies all. I'll try something out.

I think the problem you might run into using FFT is if you want a frequency that doesn't correspond to one bin. I've only been working with FFT for a short while, so take all of this with a bowl of salt, but I'm pretty sure each bin in the FFT analysis is harmonically related to the fundamental. Other frequencies are represented by combinations of bins and their phases. So filtering down to one bin will only give you a limited number of options as far as frequency choices. I imagine the amplitude might fluctuate as well, but I don't know if that's an issue for you, either.

Like I said, I'm not completely sure about all of that, so if I'm wrong, for the love of god someone correct me.

Like wise my knowledge of FFT too is shite. As far as I can disseminate, a bin is a sin wave at a certain frequency and amplitude. Something like a FFT sieve would filter out the unwanted bins. Maybe good for a random special effect. Though as Maelstorm says, i dunno how much control you get over the resulting frequency.

I think then maybe the Chebyshev distortion might be the way to go. If you have enough harmonics then it will effectively sound like 'noise', and then you can use [line] to ramp each band to 0, leaving the original. I think also this would give you more control over varying [line] rates.

The creb library has an object called [cheby~]

sounds like fun!

Thanks for the heads up on the creb object, I wasn't aware of that one.

This is my cheby patch attached (it's really just a very basic tweak of a pre-existing patch)

Start with the volume low and slowly move the slider above to the right - the further right you go, the more noise you get.

Any tweaks or refinements welcome of course... it'd be nice to get the top slider to move automatically (and smoothly).

http://www.pdpatchrepo.info/hurleur/cheby.pd

cool stuff, so I made the top slider move automatically. Is this how you mean? Sounds like a mental 80s casio synth

http://www.pdpatchrepo.info/hurleur/cheby_smoove.pd

Nice! Thanks for that... very close to the idea I had in mind. The transition from noise to sine isn't as neat as I had wanted, but I quite like the weird artfacts that doing it this way introduces.

Yeah, reminds me of a Yamaha DX100 synth I had for a while

I love how that sounds so violent, and yet ends on a peaceful tone! Thanks for sharing.

you can make the transition a little smoother by squaring the output of [line] making it exponential (or is it logarithmic?) giving it more range at the less noisy end. Also slightly low pass filtering the initial attack makes it a little less scary. [moog~] is nice for this

because i hate you you will click here

i missed this thread the first time round, and it looks like there are some great suggestions,

but to turn white noise into a sine wave, you could turn the problem on its head.

first, start with a sine wave. then apply modulations to bash it up and turn it into noise.

whatever process you followed, then reverse, and you will turn the noise back into a sine wave.

not tried this in practice, but i guess it would probably work. hardest thing is going to be getting really white noise from a sine wave.

How typically mod, turning everything into white noise

the only method i know which COULD work would be modular arithmetic everything else i remember isn't bijective.
but you would loose some bits due floating point precision
who wants to implement it?

My contrib

An fm-feedback voice

mvh

http://www.pdpatchrepo.info/hurleur/sine2noise.pd