P-modelling: Coupling K with W models - e.g. waveguide with spring

DSP, Plugin and Host development discussion.
Post Reply New Topic
RELATED
PRODUCTS

Post

I'm trying to model a brass instrument using pure physical simulation: single lip mass-spring and a bore waveguide. Has anybody had any success with coupling them? I've seen several papers on this, but there is only some higher-level math without a discrete time numerical solution. Is KW converter junction suitable for this? What quantity is coupled then?
giq

Post

Personally I just feed the pipe output into the lip simulation, and then the driver output calculated from the lip back into the pipe, while making sure that they stay in phase so that it doesn't go out of tune (so that the 1st harmonic has the same phase after being transformed by the lip+driver than what was in the pipe).

Post

MadBrain wrote:Personally I just feed the pipe output into the lip simulation, and then the driver output calculated from the lip back into the pipe, while making sure that they stay in phase so that it doesn't go out of tune (so that the 1st harmonic has the same phase after being transformed by the lip+driver than what was in the pipe).

(input pressure) <-> MASS -(pressure)<-> pipe
MASS (spring) <-> "wall"


Is it like this? no problems with spring part.. but how to calculate pipe interaction? and back? If a pipe was another spring, the force would be easily calculated by spring compression. but MASS creates pressure that is "processed" by a pipe and fed back, so it seems that some input and output variables need to be integrated..
giq

Post

I think the way your lip/reed mass behaves is not so important, as long as it's in the right phase so that the model is in tune. So it's just a kind of bandpass, that takes the complex signal from the pipe, and turns it into something simple (a sinewave) for controlling your airflow. It's not particularly important to feed the lip/reed movement back into the pipe (it's just a sinewave, which is kinda boring).

The real magic is that the flow of air depends on the position of your lips/reed in a non-linear way. This takes your sinewave and waveshapes it into the instrument's real waveshape, and it's where all your harmonics come from. So for instance, if your instrument is a bassoon, you need something that will take a sinewave (your reed position) and turn it into the kind of pulse-wave-ish-resonant-ish waveform you get from a bassoon.

Post

I know these models, like squared and clipped biquad in brass.cpp from STK. It's nice, however it lacks realism in many situations.. it's just an approximation. The waveshaped signal needs to be clipped in order to assure stability (kind of workarround?). With coupled models, the amount of energy is preserved and the flow is natural (lips vibration is no more a pure sine wave).

Nevertheless coupling iterative K-models (like mass-springs systems) and waveguides open so many possibilities..
giq

Post

How do you keep a coupled model in tune?

Post

you don't, which is desirable :) haven't said anything because when i put aerophone together i embarassingly hadn't discerned how to tune a m-s algo yet. iirc i had another go since but no bell model so no release, and also iirc staying in tune depends on the reflectivity and the k of the m-s.. yes at equal to or lower than (to some extent, maybe half an octave or so?) the frequency of the waveguide.
you come and go, you come and go. amitabha neither a follower nor a leader be tagore "where roads are made i lose my way" where there is certainty, consideration is absent.

Post Reply

Return to “DSP and Plugin Development”