Sorry the example I was looking at when I mentioned that was not really a function reference but it gets the same result. It does so by using ChucK class functionality that in other languages is called polymorphism. It is described in ChucK language documentation. http://chuck.cs.princeton.edu/doc/language See the example class Xfunc in the inheritence section. What is done is to create a base class with a standard class function (in the example called doSomething().) For each different version of function that is needed a new class is created which inherits the base class and overrides doSomething() with the new code. The "calling function" is then declared with a base class object as an argument. You can pass it any object that inherits the base class (in this case Xfunc.) When it calls the doSomething() method it invokes the overridden version from the derived class. The example doesn't create a function to do this but it shows an example of common code calling different functions from different classes. It does this by creating an array containing objects from different classes and then calling the different versions in a loop using the array. Clearly the common code in the loop could have been a function. I'm away from my ChucK machine at the moment but if anyone is interested I could try to create a simple example which does the inner work (calling a passed in function) in a function. As usual the documentation is vague but I believe the function version of the example would look something like: fun int callFunc(Xfunc @ in, int a, int b) { return in.doSomething(a,b); } // loop over the Xfunc for( 0 => int i; i < operators.cap(); i++ ) { // doSomething, potentially different for each Xfunc <<< callerFunc(operators[i], 4, 5 ) >>>; } -steve On Fri, Dec 27, 2013 at 7:03 PM, Dealga McArdle wrote:

re: Steve Morris

3) You can pass a reference to a function as an argument to another

...

function.

how? :) I haven't found how to do that anywhere.

_______________________________________________ chuck-users mailing list chuck-users@lists.cs.princeton.edu https://lists.cs.princeton.edu/mailman/listinfo/chuck-users

Of course this isn't exactly what you asked about, since it takes in an 8-bit number and you want to take in a number [0-8000], but maybe this will get you started. Maybe Steve can help with how to pass an expression as a reference, which would make it possible to change the expression on the fly. Joel On 12/27/2013 06:13 PM, Joel Matthys wrote:

On 12/27/2013 05:08 PM, Steve Morris wrote:

...

It is a mystery to many of us exactly why users are not allowed to know how to do this. Questions to this list are basically ignored except by other users. I suspect that ChucK developers don't read this list and the real point of this list is to hope users will support each other and stop bothering the developers. It is also possible that there are no real developers, that ChucK development consists of an occasional grad student who shows interest for a while then moves on to more productive endeavors.

There is a ChuGen example in the examples/extend folder; it's extremely straight-forward. I know the devs read this list, but honestly this is exactly the kind of question we as a community should be able to help with. Let's light some candles rather than curse the darkness.

Now, ChuGen takes floats in the range [-1,1] for its calculations, but you want to work on 8-bit ints, so you'll have to have functions that convert itof and ftoi. Then you can write your expression to work on those samples directly. A Phasor UGen then provides the 0-1 samples in. Try this:

-- // ChuGen // Create new UGens by performing audio-rate processing in ChucK

class MyFunc extends Chugen { 8 => int _b; // default 8 bits

fun float tick(float in) { ftoi (in, _b) => int ival; // convert in sample to int my_expression(ival) % (1 << _b) => int fval; //calculate and mod into range return (itof (fval, _b)); // convert back to float }

// your custom expression goes here fun int my_expression (int t) { return t * ( t >> 8 * ( t >>15 | t >> 8) & 20 | ( t >>19) * 5

...

...

t | t >> 3); }

// helper function to convert float [0,1] to b bits fun int ftoi (float in, int b) { return (in * (1 << b)) $ int; }

// helper function to convert b bit integer to float [-1,1] fun float itof (int in, int b) { return (in * 1.0 / (1 << (b-1))) - 1; }

// allow user to set number of bits fun void bits (int b) { b => _b; } }

// phasor powers the function with numbers ascending 0-1 Phasor p => MyFunc f => dac; 1 => p.freq;

while(true) 1::second => now;

/// END ///

Joel _______________________________________________ chuck-users mailing list chuck-users@lists.cs.princeton.edu https://lists.cs.princeton.edu/mailman/listinfo/chuck-users

See my post about polymorphism. To "pass in an expression in on the fly" you need to wrap the function in a polymorphic class and "on the fly" pass a reference to that class instead. For example if you are doing this in a ChuGen you can create a class variable to hold a reference to the "current" expression object. At the appropriate time you can assign a new class to the reference variable in much the same way you would change the gain or frequency on the fly. The tick() method would call the current function through the reference variable. You can then put the common code directly in the ChuGen and code you want to change at run time in the referenced object. This is all simpler if all classes are defined in the same file and thread but I believe that if you are clever the new code doesn't even need to be defined when the ChuGen is started. All it needs is for the base class to be defined. You could connect and start up the ChuGen and then weeks later write new code and put it into the reference variable of the running ChuGen. The tricky part is how to pass the new object from a different thread. This requires using static class variables which is a subject for a different post. -steve PS all the helper functions you can imagine are probably already written and available in the LiCK Library for ChucK. If you don't have it already you should get it immediately. Half the things I was originally planning of writing in ChucK I found already implemented in LiCK. It is also a great thing to read if you want to see how complex things can be done in ChuCK. On Fri, Dec 27, 2013 at 7:16 PM, Joel Matthys wrote:

Of course this isn't exactly what you asked about, since it takes in an 8-bit number and you want to take in a number [0-8000], but maybe this will get you started.

Maybe Steve can help with how to pass an expression as a reference, which would make it possible to change the expression on the fly.

Joel

On 12/27/2013 06:13 PM, Joel Matthys wrote:

...

On 12/27/2013 05:08 PM, Steve Morris wrote:

...

It is a mystery to many of us exactly why users are not allowed to know how to do this. Questions to this list are basically ignored except by other users. I suspect that ChucK developers don't read this list and the real point of this list is to hope users will support each other and stop bothering the developers. It is also possible that there are no real developers, that ChucK development consists of an occasional grad student who shows interest for a while then moves on to more productive endeavors.

There is a ChuGen example in the examples/extend folder; it's extremely straight-forward. I know the devs read this list, but honestly this is exactly the kind of question we as a community should be able to help with. Let's light some candles rather than curse the darkness.

Now, ChuGen takes floats in the range [-1,1] for its calculations, but you want to work on 8-bit ints, so you'll have to have functions that convert itof and ftoi. Then you can write your expression to work on those samples directly. A Phasor UGen then provides the 0-1 samples in. Try this:

-- // ChuGen // Create new UGens by performing audio-rate processing in ChucK

class MyFunc extends Chugen { 8 => int _b; // default 8 bits

fun float tick(float in) { ftoi (in, _b) => int ival; // convert in sample to int my_expression(ival) % (1 << _b) => int fval; //calculate and mod into range return (itof (fval, _b)); // convert back to float }

// your custom expression goes here fun int my_expression (int t) { return t * ( t >> 8 * ( t >>15 | t >> 8) & 20 | ( t >>19) * 5 >> t | t >> 3); }

// helper function to convert float [0,1] to b bits fun int ftoi (float in, int b) { return (in * (1 << b)) $ int; }

// helper function to convert b bit integer to float [-1,1] fun float itof (int in, int b) { return (in * 1.0 / (1 << (b-1))) - 1; }

// allow user to set number of bits fun void bits (int b) { b => _b; } }

// phasor powers the function with numbers ascending 0-1 Phasor p => MyFunc f => dac; 1 => p.freq;

while(true) 1::second => now;

/// END ///

Joel _______________________________________________ chuck-users mailing list chuck-users@lists.cs.princeton.edu https://lists.cs.princeton.edu/mailman/listinfo/chuck-users

_______________________________________________ chuck-users mailing list chuck-users@lists.cs.princeton.edu https://lists.cs.princeton.edu/mailman/listinfo/chuck-users

Hi again, these days i been experimenting with the code Joel posted, i discovered some things and got some questions. where is the time of the waveform defined? i experimented changing all values and also with many expression and the length of the waveform remains the same ( there is a waveform that loops for ever) . do anybody have an idea where the length of the waveform is defined and how to change it? Other thing i noticed is that changing the values in the expressions changes the timbre ( or texture ? ) of the sound in a subtle way while the structure remains the same. So i was thinking that instead of having values in the expression, Is it possible to have an oscilator or brown noise generator? .......having a brown noise generator instead of a value will allow that the texture of the sound will change all the time. Other thing i noticed is that the paper is posted before and the algorithm for generating this kind of functions is based on the implementation from the links Michael posted before. If you listen those examples you will notice that the sound changes all the time, there is no repetitiveness , it creates sound structures at different time scale and this is the potential in this way of sonifying expression. any idea why Joels implementation only generate a sound withouth variability as the examples posted by Michael? and my last question: I been working with other programming languages for sound as supercollider and there are different ways of generating waveforms from functions, basically the way i was doing this is that first i create the function and then i call the function that creates the whole waveform and then i just read my waveform but i see in the code posted by Joel that it works in a different way : first i generate the function and then i generate the waveform at the same time I generate the sound calling the function each sample . I was wondering is this something specific to chuck? I mean calling the function each sample while it generates the sound.... is this approach possible with other programming languages? 2013/12/29, Ronni Montoya :

hey, i found a link to the paper, i hope you enjoy it:

http://cilab.math.upatras.gr/mikeagn/sites/cilab.math.upatras.gr.mikeagn/fil...

2013/12/28, Michael Heuer :

...

Hello Ronni,

Might you have a link to the paper you are referring to? Something is lost to me in the specification of the algorithm.

I am reminded of this

http://createdigitalmusic.com/2011/10/entire-musical-compositions-made-from-...

http://countercomplex.blogspot.com/2011/10/algorithmic-symphonies-from-one-l...

http://countercomplex.blogspot.com/2011/10/some-deep-analysis-of-one-line-mu...

Cheers,

michael

On Fri, Dec 27, 2013 at 1:46 AM, Ronni Montoya wrote:

...

Hi list, i been reading a paper that talks about an approach for generating sound that uses functions to directly shape waveforms. The function receives "expresions" as argument and it generates different kind of 8 bit waveforms from the expressions.

An expression can look like this:

f(t) = t * ( t >> 8 * ( t >>15 | t >> 8) & 20 | ( t >>19) * 5 >> t | t

...

...

3))

The paper describes the algorithm in this way:

algorithm 1 : Construction of an 8-bit waveform of 8000Hz sample rate:

The function has 2 inputs :

Input 1) a functional expresion f(t) Input 2 ) time durations in seconds (d) Output: the waveform of an audio signal with seconds duration

for t = 1 to d8000 do if f(t) ==NaN then q(t) <-- 0 else q(t) <-- mod(f(t)), 256) end if s(t) <-- 2 q(t)/255 - 1 end for

If its possible to implement this approach for generating sound in chuck??

cheers

R. _______________________________________________ chuck-users mailing list chuck-users@lists.cs.princeton.edu https://lists.cs.princeton.edu/mailman/listinfo/chuck-users

---

chuck-users mailing list chuck-users@lists.cs.princeton.edu https://lists.cs.princeton.edu/mailman/listinfo/chuck-users