Alex Pojda

I'm synthesizing a steam whistle (like the ones in steam locomotives or calliope organs).

I derived a quasi physical model from Perry Cook's slide flute where all parameters that shape the sound are controlled with just one slider (like the handle you pull in a locomotive).

My patch is functional and sounds somewhat like a steam whistle in a certain narrow range of the slider but something seems to be wrong in my gen signal flow.

Essentially the only thing that needs to be inverted is the pitch change with increasing slider value. Right now it decreases while the noise level increases and I can't get this right, can anyone take a look?

I'm synthesizing a steam whistle (like the ones in steam locomotives or calliope organs).
I derived a quasi physical model from Perry Cook's slide flute where all parameters that shape the sound are controlled with just one slider (like the handle you pull in a locomotive).
My patch is functional and sounds somewhat like a steam whistle in a certain narrow range of the slider but something seems to be wrong in my gen signal flow.
Essentially the only thing that needs to be inverted is the pitch change with increasing slider value. Right now it decreases while the noise level increases and I can't get this right, can anyone take a look?


waveguide-synthesis-modeling-a-steamwhistle

Oh and btw. I don't get an oscillation when i use the cubic nonlinearity as proposed by Cook! I use tanh instead and it works fine, but aren't these functions quite similar as waveshapers?!

Are you pulling the idea from Real Sound Synthesis for Interactive Applications by Perry Cook?

Yes, there is a gen~ example implementation from that book for the slide flute in Max 7 which i dissected and rewired.

The problem is that i don't know how to get the "bell delay" value to decrease with an increase of the slider in my patch. With the slide flute model that behaviour is normal since an increase in slide length lowers the pitch (which has a separate slider in his patch). I need to reverse that somehow so that my "bell delay" time somehow ranges between let's say 44 samples and 100 samples, but i can't figure out an equation to transform the output from the noisecontrol into the needed sample delay values.

Yes, there is a gen~ example implementation from that book for the slide flute in Max 7 which i dissected and rewired.
The problem is that i don't know how to get the "bell delay" value to decrease with an increase of the slider in my patch. With the slide flute model that behaviour is normal since an increase in slide length lowers the pitch (which has a separate slider in his patch). I need to reverse that somehow so that my "bell delay" time somehow ranges between let's say 44 samples and 100 samples, but i can't figure out an equation to transform the output from the noisecontrol into the needed sample delay values.


How about using scale? For example, scale 0. 1. 100. 44.

If you need exponential scaling, try scale 0. 1. 100. 44. 2 @classic 0. (where 2 is the exponent)

How about using scale?  For example, scale 0. 1. 100. 44.

If you need exponential scaling, try scale 0. 1. 100. 44. 2 @classic 0. (where 2 is the exponent)


Thanks for the tip Peter, but it doesn't seem to work. When i use scale the pitch remains constant... What puzzles me is that when set an output in gen as a measurement point to monitor values from the outputs of each branching point, the bell delay never receives zero, even when the slider is all down. I don't know where it's values are coming from. Does that have do to with tanh?

Thanks for the tip Peter,  but it doesn't seem to work. When i use scale the pitch remains constant... What puzzles me is that when set an output in gen as a measurement point to monitor values from the outputs of each branching point, the bell delay never receives zero, even when the slider is all down. I don't know where it's values are coming from. Does that have do to with tanh?


Have you tried varying the other delay? That did it for me.

Have you tried varying the other delay?  That did it for me.


can you post how you did that in the patch? i used scale on the other delay like you proposed but it only causes the whistle to sound like an overblown flute...

Update: I tried reversing the pressure slider for the white noise. Now it starts with a low pitch and rises as i raise the slider, but the white noise behaves the same, means it fades away as i pull the fader up... it's a twisted riddle!

[Waveguide Synthesis] Modeling a Steamwhistle