I'm working on a project which involves porting some Csound code from 
the Csound Book to MaxMSP, and I can't figure out which MSP object will 
perform the same function as Csound's tone opcode.  I know I can do it 
with biquad~, but I'm not sure how to figure out the correct 
coefficients.  Any suggestions will be very helpful.

Thanks,
Rich

On Oct 9, 2005, at 9:22 PM, Richard M. Otero wrote:

> I'm working on a project which involves porting some Csound code from 
> the Csound Book to MaxMSP, and I can't figure out which MSP object 
> will perform the same function as Csound's tone opcode.  I know I can 
> do it with biquad~, but I'm not sure how to figure out the correct 
> coefficients.  Any suggestions will be very helpful.
>
> Thanks,
> Rich
>

Hi Rich,

According to the Max/MSP help file for [reson~]:

reson~ implements the following filter equation:

y[n] = gain * (x[n] - r * x[n-2]) + c1 * y[n-1] + c2 * y[n-2],  ...

where r, c1, and c2 are parameters calculated from the input center 
frequency and Q.

Whereas from the Csound Manual:

Tone is a 1 term IIR filter. Its formula is:

y[n] = c1 * x[n] + c2 * y[n-1] + c3 * y[n-2] ...

So other than the added control of Q (inverse of ƒc) it seems to be 
implementing a similar filter algorithm, perhaps this could be suitable 
for your needs?


-David

David Akbari wrote:

> Tone is a 1 term IIR filter. Its formula is:
> 
> y[n] = c1 * x[n] + c2 * y[n-1] + c3 * y[n-2] ...

tone is actually

   y[n] = c1 * x[n] + c2 * y[n-1]

where c1 and c2 are calculated as follows:

   b = 2 - cos(khp * 2 * PI / sr)
   c2 = b - sqrt(b * b - 1)
   c1 = 1 - c2

khp is the cutoff frequency, sr is the sample rate, and
b is a temporary variable

Is this in the csound manual? If so, it must be
corrected as it is wrong. The Filter equation
quoted is second-order and (depending on c1,c2, c3)
implements a resonator.

A 1st-order low-pass IIR that implements tone is:

[cf is cuttoff frequency]

a = 2 - cos(2*pi*cf/sr)
c2 = sqrt(a^2 -1) - a
c1 = 1 + c2

y[n] = c1*x[n] - c2*y[n-1]

[for a high-pass 'flip', just use -c2 instead of c2 in the equations
above].

Victor


>Tone is a 1 term IIR filter. Its formula is:
>
>y[n] = c1 * x[n] + c2 * y[n-1] + c3 * y[n-2] ...
>
>So other than the added control of Q (inverse of ƒc) it seems to be 
>implementing a similar filter algorithm, perhaps this could be suitable 
>for your needs?
>
>
>-David
>--
>Send bugs reports to this list.
>To unsubscribe, send email to csound-unsubscribe@lists.bath.ac.uk

Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 

Thanks, Victor, Istvan, and Dave.

Just to clarify for my own sake:

y[n] = c1*x[n] -c2*y[n-1] is a first-order lowpass IIR filter (Victor's 
example)

and

y[n] = c1*x[n]+c2*y[n-1] is a first-order highpass IIR filter (Istvan's 
example via Victor's explanation of the use of -c2 instead of c2).

I'm not sure what Dave's example is:

y[n] = c1*x[n]+c2*y[n-1]+c3*y[n-2] ...

A higher-order highpass IIR filter?  Is each occurence of the c 
variable a pole?  I hope Dr. B doesn't see this thread because I 
probably should have memorized all this when I took his DSP class.  
I'll have to consult my notes...

Thanks again for your help,
Rich

On Oct 10, 2005, at 4:19 AM, Victor Lazzarini wrote:

> Is this in the csound manual? If so, it must be
> corrected as it is wrong. The Filter equation
> quoted is second-order and (depending on c1,c2, c3)
> implements a resonator.
>
> A 1st-order low-pass IIR that implements tone is:
>
> [cf is cuttoff frequency]
>
> a = 2 - cos(2*pi*cf/sr)
> c2 = sqrt(a^2 -1) - a
> c1 = 1 + c2
>
> y[n] = c1*x[n] - c2*y[n-1]
>
> [for a high-pass 'flip', just use -c2 instead of c2 in the equations
> above].
>
> Victor
>
>
>> Tone is a 1 term IIR filter. Its formula is:
>>
>> y[n] = c1 * x[n] + c2 * y[n-1] + c3 * y[n-2] ...
>>
>> So other than the added control of Q (inverse of ƒc) it seems to be 
>> implementing a similar filter algorithm, perhaps this could be 
>> suitable for your needs?
>>
>>
>> -David
>> --
>> Send bugs reports to this list.
>> To unsubscribe, send email to csound-unsubscribe@lists.bath.ac.uk
>
> Victor Lazzarini
> Music Technology Laboratory
> Music Department
> National University of Ireland, Maynooth
> --
> Send bugs reports to this list.
> To unsubscribe, send email to csound-unsubscribe@lists.bath.ac.uk
>

On Oct 10, 2005, at 11:45 PM, Richard M. Otero wrote:

> I'm not sure what Dave's example is:
>
> y[n] = c1*x[n]+c2*y[n-1]+c3*y[n-2] ...
>

OMG I just realized the error ... I was thinking of [lores~] when i 
typed that!~! They say your mental capacity drops by over 50% and 
declines rapidly after you have been up for 17+ hours.

In any event your best bet on implementing it in MSP would be probably 
to use the [expr~] object. Although I'm not certain it works well in 
MaxMSP?? (I know it's great in Pd!) That way you can ensure that it is 
implemented exactly as it is in Csound.

> I hope Dr. B doesn't see this thread because I probably should have 
> memorized all this when I took his DSP class.
>

LOL I know!! ... *hides*


-David

yes, but remember that in that case

c1 = 1 - c2

(as c2 from the lp example is now -c2). So you can
actually use the same filter equation, but just change
c1 and c2

y[n] = c1*x[n] - c2*y[n-1]

intermediary variable:
a = 2 - cos(2pi*cf/sr)

lp case: c2 = sqrt(a^2-1) - a,  c1 = 1 + c2
hp case: c2= a - sqrt(a^2-1), c1 = 1 - c2

I hope it's clear now. Dave's example generally implements
a 2nd order BP filter, a reson if you use

c2 = 2Rcos(2*pi*fc/sr), c3 = -R^2

where R is the radius (calculated from the BW) and
fc is the centre frequency.

Victor


At 04:45 11/10/2005, you wrote:
>y[n] = c1*x[n]+c2*y[n-1] is a first-order highpass IIR filter (Istvan's 
>example via Victor's explanation of the use of -c2 instead of c2).

Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 

Richard M. Otero wrote:

> y[n] = c1*x[n] -c2*y[n-1] is a first-order lowpass IIR filter (Victor's 
> example)
> 
> and
> 
> y[n] = c1*x[n]+c2*y[n-1] is a first-order highpass IIR filter (Istvan's 
> example via Victor's explanation of the use of -c2 instead of c2).

My example actually behaves in the same way, as even though +c2 is
used instead of -c2, c2 is calculated with a different sign. In fact,
I took the code directly from the tone opcode itself.