Dear List,

I'm trying to teach myself about Physical Modeling Synthesis at the 
moment. In doing so I'm slowly getting to grips with filters.

I've decided, as an exercise, to have a go at modeling a banjo. I 
recorded the sound of tapping the banjo bridge with the strings damped, 
and I identified some resonances of the banjo's head and pot (using 
transforms in snd). I've started by trying various band-pass filters to 
model the resonances, and playing the "pluck" opcodes through them. It 
sounds OK... but not great!

One problem I'm facing is that many of the filters take very different 
kinds of arguments and ranges. This makes them difficult to compare as 
they're not drop-in replacements for each other. Are there any 
particularly appropriate filter opcodes for this situation? Or any 
advice, techniques or tricks for doing this kind of thing would be 
greatly appreciated!

Many thanks,

Joe

-- 
Joseph Sanger

Blueberry Eikaiwa
www.bbeikaiwa.com
joe@bbeikaiwa.com

Perhaps if you share your CSD with us, we might be able to
give an opinion on it.

Generally speaking, resonators are your best bet when attempting
this kind of thing. Another thing you could do is try to get an
impulse response of your system and then use pconvolve to
try and reproduce it.

Regards

Victor
----- Original Message ----- 
From: "Joseph Sanger" 
To: 
Sent: Thursday, April 30, 2009 2:43 PM
Subject: [Csnd] filters for physical modeling


> Dear List,
>
> I'm trying to teach myself about Physical Modeling Synthesis at the 
> moment. In doing so I'm slowly getting to grips with filters.
>
> I've decided, as an exercise, to have a go at modeling a banjo. I recorded 
> the sound of tapping the banjo bridge with the strings damped, and I 
> identified some resonances of the banjo's head and pot (using transforms 
> in snd). I've started by trying various band-pass filters to model the 
> resonances, and playing the "pluck" opcodes through them. It sounds OK... 
> but not great!
>
> One problem I'm facing is that many of the filters take very different 
> kinds of arguments and ranges. This makes them difficult to compare as 
> they're not drop-in replacements for each other. Are there any 
> particularly appropriate filter opcodes for this situation? Or any advice, 
> techniques or tricks for doing this kind of thing would be greatly 
> appreciated!
>
> Many thanks,
>
> Joe
>
> -- 
> Joseph Sanger
>
> Blueberry Eikaiwa
> www.bbeikaiwa.com
> joe@bbeikaiwa.com
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

Thank you, when I get a chance I'll tidy it up a bit and put some 
comments on, then I'll post it and see what you think. It's 
embarrassingly messy at the moment.

Yours,

Joe

victor wrote:
> Perhaps if you share your CSD with us, we might be able to
> give an opinion on it.
> 
> Generally speaking, resonators are your best bet when attempting
> this kind of thing. Another thing you could do is try to get an
> impulse response of your system and then use pconvolve to
> try and reproduce it.
> 
> Regards
> 
> Victor
> ----- Original Message ----- From: "Joseph Sanger" 
> To: 
> Sent: Thursday, April 30, 2009 2:43 PM
> Subject: [Csnd] filters for physical modeling
> 
> 
>> Dear List,
>>
>> I'm trying to teach myself about Physical Modeling Synthesis at the 
>> moment. In doing so I'm slowly getting to grips with filters.
>>
>> I've decided, as an exercise, to have a go at modeling a banjo. I 
>> recorded the sound of tapping the banjo bridge with the strings 
>> damped, and I identified some resonances of the banjo's head and pot 
>> (using transforms in snd). I've started by trying various band-pass 
>> filters to model the resonances, and playing the "pluck" opcodes 
>> through them. It sounds OK... but not great!
>>
>> One problem I'm facing is that many of the filters take very different 
>> kinds of arguments and ranges. This makes them difficult to compare as 
>> they're not drop-in replacements for each other. Are there any 
>> particularly appropriate filter opcodes for this situation? Or any 
>> advice, techniques or tricks for doing this kind of thing would be 
>> greatly appreciated!
>>
>> Many thanks,
>>
>> Joe
>>
>> -- 
>> Joseph Sanger
>>
>> Blueberry Eikaiwa
>> www.bbeikaiwa.com
>> joe@bbeikaiwa.com
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>> "unsubscribe csound" 
> 
> 
> 
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound"
> 
> 

-- 
Joseph Sanger

Blueberry Eikaiwa
www.bbeikaiwa.com
joe@bbeikaiwa.com

I am looking for ideas on how I could "excite" a signal
and add more character to it. The first thing that 
pops into my mind is to run it through some resonator.

Does anyone have any examples out there that do this?
What opcodes would I use to create a resonator?


Thanks,
Anthony

I like to excite my signals by taking them out to a nice dinner but
you might like to try using some comb filters, it really gets them
going!

On 01/05/2009, Anthony Palomba  wrote:
> I am looking for ideas on how I could "excite" a signal
>  and add more character to it. The first thing that pops into my mind is to
> run it through some resonator.
>
>  Does anyone have any examples out there that do this?
>  What opcodes would I use to create a resonator?
>
>
>  Thanks,
>  Anthony
>
>
>  Send bugs reports to this list.
>  To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
> csound"
>

> I am looking for ideas on how I could "excite" a signal
> and add more character to it. The first thing that
> pops into my mind is to run it through some resonator.
>
> Does anyone have any examples out there that do this?
> What opcodes would I use to create a resonator?
>
>
> Thanks,

ome toughts:

reverb etc
repuck
streson
reson
nlfilt

Streson is a good bet, gives a nice "piano cage" effect.
----- Original Message ----- 
From: "Anthony Palomba" 
To: 
Sent: Friday, May 01, 2009 1:47 PM
Subject: [Csnd] resonator examples...


>I am looking for ideas on how I could "excite" a signal
> and add more character to it. The first thing that pops into my mind is to 
> run it through some resonator.
>
> Does anyone have any examples out there that do this?
> What opcodes would I use to create a resonator?
>
>
> Thanks,
> Anthony
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

I guess band-pass filters may not be ideal for this. Resonances are 
better modeled with resonance filters, the only opcode I know that does 
it is mode (rezzy may also be good, but the manual says it is "created 
empirically", I don't know what it does). The opcode reson is a 
band-pass filter, so its name is actually misleading.

Joseph Sanger wrote:
> Dear List,
>
> I'm trying to teach myself about Physical Modeling Synthesis at the 
> moment. In doing so I'm slowly getting to grips with filters.
>
> I've decided, as an exercise, to have a go at modeling a banjo. I 
> recorded the sound of tapping the banjo bridge with the strings 
> damped, and I identified some resonances of the banjo's head and pot 
> (using transforms in snd). I've started by trying various band-pass 
> filters to model the resonances, and playing the "pluck" opcodes 
> through them. It sounds OK... but not great!
>
> One problem I'm facing is that many of the filters take very different 
> kinds of arguments and ranges. This makes them difficult to compare as 
> they're not drop-in replacements for each other. Are there any 
> particularly appropriate filter opcodes for this situation? Or any 
> advice, techniques or tricks for doing this kind of thing would be 
> greatly appreciated!
>
> Many thanks,
>
> Joe
>

-- 
  Mark
  _________________________________________
  When you get lemons, you make lemonade.
  When you get hardware, you make software.

Physical Models is not my area, but as far as I know, resonances
are modelled by resonators, as discussed in

http://www.music.mcgill.ca/~gary/307/week9/node4.html

That's your reson filter alright.

Victor

----- Original Message ----- 
From: "Mark Van Peteghem" 
To: 
Sent: Friday, May 01, 2009 8:29 PM
Subject: [Csnd] Re: filters for physical modeling


>I guess band-pass filters may not be ideal for this. Resonances are better 
>modeled with resonance filters, the only opcode I know that does it is mode 
>(rezzy may also be good, but the manual says it is "created empirically", I 
>don't know what it does). The opcode reson is a band-pass filter, so its 
>name is actually misleading.
>
> Joseph Sanger wrote:
>> Dear List,
>>
>> I'm trying to teach myself about Physical Modeling Synthesis at the 
>> moment. In doing so I'm slowly getting to grips with filters.
>>
>> I've decided, as an exercise, to have a go at modeling a banjo. I 
>> recorded the sound of tapping the banjo bridge with the strings damped, 
>> and I identified some resonances of the banjo's head and pot (using 
>> transforms in snd). I've started by trying various band-pass filters to 
>> model the resonances, and playing the "pluck" opcodes through them. It 
>> sounds OK... but not great!
>>
>> One problem I'm facing is that many of the filters take very different 
>> kinds of arguments and ranges. This makes them difficult to compare as 
>> they're not drop-in replacements for each other. Are there any 
>> particularly appropriate filter opcodes for this situation? Or any 
>> advice, techniques or tricks for doing this kind of thing would be 
>> greatly appreciated!
>>
>> Many thanks,
>>
>> Joe
>>
>
> -- 
>  Mark
>  _________________________________________
>  When you get lemons, you make lemonade.
>  When you get hardware, you make software.
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

Hi there Victor, Mark and List,

Thanks for your replies, I'm at work now so I'll check the link later.

Here is the csd of the banjo "so far". I'm fairly happy, but it's not 
finished yet! After experimenting with various filters I ended up with 
resonr - mode seemed to isolate the frequency too much.

It seems that my problem has shifted to the string sound, rather than 
the resonating body. I haven't really started experimenting properly 
yet, but I want to model each of the five strings separately to allow 
for resonance and coupling between the strings (an important part of the 
banjo's sound!). I also want to model the banjo picks, and pick 
positions, so I need to really understand waveguide strings.

Any advice would be welcome,

Joe

victor wrote:
> Physical Models is not my area, but as far as I know, resonances
> are modelled by resonators, as discussed in
> 
> http://www.music.mcgill.ca/~gary/307/week9/node4.html
> 
> That's your reson filter alright.
> 
> Victor
> 
> ----- Original Message ----- From: "Mark Van Peteghem" 
> 
> To: 
> Sent: Friday, May 01, 2009 8:29 PM
> Subject: [Csnd] Re: filters for physical modeling
> 
> 
>> I guess band-pass filters may not be ideal for this. Resonances are 
>> better modeled with resonance filters, the only opcode I know that 
>> does it is mode (rezzy may also be good, but the manual says it is 
>> "created empirically", I don't know what it does). The opcode reson is 
>> a band-pass filter, so its name is actually misleading.
>>
>> Joseph Sanger wrote:
>>> Dear List,
>>>
>>> I'm trying to teach myself about Physical Modeling Synthesis at the 
>>> moment. In doing so I'm slowly getting to grips with filters.
>>>
>>> I've decided, as an exercise, to have a go at modeling a banjo. I 
>>> recorded the sound of tapping the banjo bridge with the strings 
>>> damped, and I identified some resonances of the banjo's head and pot 
>>> (using transforms in snd). I've started by trying various band-pass 
>>> filters to model the resonances, and playing the "pluck" opcodes 
>>> through them. It sounds OK... but not great!
>>>
>>> One problem I'm facing is that many of the filters take very 
>>> different kinds of arguments and ranges. This makes them difficult to 
>>> compare as they're not drop-in replacements for each other. Are there 
>>> any particularly appropriate filter opcodes for this situation? Or 
>>> any advice, techniques or tricks for doing this kind of thing would 
>>> be greatly appreciated!
>>>
>>> Many thanks,
>>>
>>> Joe
>>>
>>
>> -- 
>>  Mark
>>  _________________________________________
>>  When you get lemons, you make lemonade.
>>  When you get hardware, you make software.
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>> "unsubscribe csound" 
> 
> 
> 
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound"
> 
> 

-- 
Joseph Sanger

Blueberry Eikaiwa
www.bbeikaiwa.com
joe@bbeikaiwa.com

That page is confusing to me, it talks about resonators but the graph is 
from a bandpass filter (0 gain at freq. 0, i.e. real poles). Resonance 
filters with complex poles have a gain that is not 0 when the frequency 
is zero, see . Maybe the word 
resonance is used differently in physics and digital music (I have a 
background in physics). And maybe one can use bandpass filters to make 
it sound good, although physics tells us to use resonance.

I looked a little more at the reson opcode, and I'm confused here as 
well. The CSound manual says that reson is a second order _resonant_ 
filter (i.e. complex poles). But 
 (at about two thirds of the 
page) says it is a _band-pass_ filter; the graph clearly is that of a 
filter with real poles.

In my first year at the university I had a course on vibrations and 
resonances, and it never mentioned band-pass filters, only resonance 
filters.

victor wrote:
> Physical Models is not my area, but as far as I know, resonances
> are modelled by resonators, as discussed in
>
> http://www.music.mcgill.ca/~gary/307/week9/node4.html
>
> That's your reson filter alright.
>
> Victor
>
> ----- Original Message ----- From: "Mark Van Peteghem" 
> 
> To: 
> Sent: Friday, May 01, 2009 8:29 PM
> Subject: [Csnd] Re: filters for physical modeling
>
>
>> I guess band-pass filters may not be ideal for this. Resonances are 
>> better modeled with resonance filters, the only opcode I know that 
>> does it is mode (rezzy may also be good, but the manual says it is 
>> "created empirically", I don't know what it does). The opcode reson 
>> is a band-pass filter, so its name is actually misleading.
>>
>> Joseph Sanger wrote:
>>> Dear List,
>>>
>>> I'm trying to teach myself about Physical Modeling Synthesis at the 
>>> moment. In doing so I'm slowly getting to grips with filters.
>>>
>>> I've decided, as an exercise, to have a go at modeling a banjo. I 
>>> recorded the sound of tapping the banjo bridge with the strings 
>>> damped, and I identified some resonances of the banjo's head and pot 
>>> (using transforms in snd). I've started by trying various band-pass 
>>> filters to model the resonances, and playing the "pluck" opcodes 
>>> through them. It sounds OK... but not great!
>>>
>>> One problem I'm facing is that many of the filters take very 
>>> different kinds of arguments and ranges. This makes them difficult 
>>> to compare as they're not drop-in replacements for each other. Are 
>>> there any particularly appropriate filter opcodes for this 
>>> situation? Or any advice, techniques or tricks for doing this kind 
>>> of thing would be greatly appreciated!
>>>
>>> Many thanks,
>>>
>>> Joe
>>>
>>
>> -- 
>>  Mark
>>  _________________________________________
>>  When you get lemons, you make lemonade.
>>  When you get hardware, you make software.
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>> "unsubscribe csound" 
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
> "unsubscribe csound"
>
>

-- 
  Mark
  _________________________________________
  When you get lemons, you make lemonade.
  When you get hardware, you make software.

Hi Joseph,

If mode isolates a frequency too much, you should lower the iq 
parameter. I also noticed that you don't scale the signals created with 
resonr, it may help to make it something like

aexcite	= (iscale1*afilt1+iscale2*afilt2+iscale3*afilt3+...)


Mark

Joseph Sanger wrote:
> Hi there Victor, Mark and List,
>
> Thanks for your replies, I'm at work now so I'll check the link later.
>
> Here is the csd of the banjo "so far". I'm fairly happy, but it's not 
> finished yet! After experimenting with various filters I ended up with 
> resonr - mode seemed to isolate the frequency too much.
>
> It seems that my problem has shifted to the string sound, rather than 
> the resonating body. I haven't really started experimenting properly 
> yet, but I want to model each of the five strings separately to allow 
> for resonance and coupling between the strings (an important part of 
> the banjo's sound!). I also want to model the banjo picks, and pick 
> positions, so I need to really understand waveguide strings.
>
> Any advice would be welcome,
>
> Joe
>
> victor wrote:
>> Physical Models is not my area, but as far as I know, resonances
>> are modelled by resonators, as discussed in
>>
>> http://www.music.mcgill.ca/~gary/307/week9/node4.html
>>
>> That's your reson filter alright.
>>
>> Victor
>>
>> ----- Original Message ----- From: "Mark Van Peteghem" 
>> 
>> To: 
>> Sent: Friday, May 01, 2009 8:29 PM
>> Subject: [Csnd] Re: filters for physical modeling
>>
>>
>>> I guess band-pass filters may not be ideal for this. Resonances are 
>>> better modeled with resonance filters, the only opcode I know that 
>>> does it is mode (rezzy may also be good, but the manual says it is 
>>> "created empirically", I don't know what it does). The opcode reson 
>>> is a band-pass filter, so its name is actually misleading.
>>>
>>> Joseph Sanger wrote:
>>>> Dear List,
>>>>
>>>> I'm trying to teach myself about Physical Modeling Synthesis at the 
>>>> moment. In doing so I'm slowly getting to grips with filters.
>>>>
>>>> I've decided, as an exercise, to have a go at modeling a banjo. I 
>>>> recorded the sound of tapping the banjo bridge with the strings 
>>>> damped, and I identified some resonances of the banjo's head and 
>>>> pot (using transforms in snd). I've started by trying various 
>>>> band-pass filters to model the resonances, and playing the "pluck" 
>>>> opcodes through them. It sounds OK... but not great!
>>>>
>>>> One problem I'm facing is that many of the filters take very 
>>>> different kinds of arguments and ranges. This makes them difficult 
>>>> to compare as they're not drop-in replacements for each other. Are 
>>>> there any particularly appropriate filter opcodes for this 
>>>> situation? Or any advice, techniques or tricks for doing this kind 
>>>> of thing would be greatly appreciated!
>>>>
>>>> Many thanks,
>>>>
>>>> Joe
>>>>
>>>
>>> -- 
>>>  Mark
>>>  _________________________________________
>>>  When you get lemons, you make lemonade.
>>>  When you get hardware, you make software.
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>> "unsubscribe csound" 
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>> "unsubscribe csound"
>>
>>
>

-- 
  Mark
  _________________________________________
  When you get lemons, you make lemonade.
  When you get hardware, you make software.

Csound's reson is a 2-order all-pole resonator, with two poles that are
complex-conjugate of each other at  Rexp(iw) and Rexp(-iw) (w = 2*pi*fc/fs).

The filter in the link I sent is a common variation on the reson theme,
with zeros placed at DC/Nyquist (so not all-pole, but poles+zeros).
That corresponds to resonz in Csound. For more details you can
follow the references in the csound manual page.

I am not sure I follow what you are saying about bandpass having
complex poles. IIR Filters that are 2-order or higher will exhibit
complex-valued poles, regardless of the amplitude response shape
(you can make them lowpass,bandpass,highpass).

Victor

----- Original Message ----- 
From: "Mark Van Peteghem" 
To: 
Sent: Saturday, May 02, 2009 6:50 AM
Subject: [Csnd] Re: Re: Re: filters for physical modeling


> That page is confusing to me, it talks about resonators but the graph is 
> from a bandpass filter (0 gain at freq. 0, i.e. real poles). Resonance 
> filters with complex poles have a gain that is not 0 when the frequency is 
> zero, see . Maybe the word 
> resonance is used differently in physics and digital music (I have a 
> background in physics). And maybe one can use bandpass filters to make it 
> sound good, although physics tells us to use resonance.
>
> I looked a little more at the reson opcode, and I'm confused here as well. 
> The CSound manual says that reson is a second order _resonant_ filter 
> (i.e. complex poles). But  (at 
> about two thirds of the page) says it is a _band-pass_ filter; the graph 
> clearly is that of a filter with real poles.
>
> In my first year at the university I had a course on vibrations and 
> resonances, and it never mentioned band-pass filters, only resonance 
> filters.
>
> victor wrote:
>> Physical Models is not my area, but as far as I know, resonances
>> are modelled by resonators, as discussed in
>>
>> http://www.music.mcgill.ca/~gary/307/week9/node4.html
>>
>> That's your reson filter alright.
>>
>> Victor
>>
>> ----- Original Message ----- From: "Mark Van Peteghem" 
>> 
>> To: 
>> Sent: Friday, May 01, 2009 8:29 PM
>> Subject: [Csnd] Re: filters for physical modeling
>>
>>
>>> I guess band-pass filters may not be ideal for this. Resonances are 
>>> better modeled with resonance filters, the only opcode I know that does 
>>> it is mode (rezzy may also be good, but the manual says it is "created 
>>> empirically", I don't know what it does). The opcode reson is a 
>>> band-pass filter, so its name is actually misleading.
>>>
>>> Joseph Sanger wrote:
>>>> Dear List,
>>>>
>>>> I'm trying to teach myself about Physical Modeling Synthesis at the 
>>>> moment. In doing so I'm slowly getting to grips with filters.
>>>>
>>>> I've decided, as an exercise, to have a go at modeling a banjo. I 
>>>> recorded the sound of tapping the banjo bridge with the strings damped, 
>>>> and I identified some resonances of the banjo's head and pot (using 
>>>> transforms in snd). I've started by trying various band-pass filters to 
>>>> model the resonances, and playing the "pluck" opcodes through them. It 
>>>> sounds OK... but not great!
>>>>
>>>> One problem I'm facing is that many of the filters take very different 
>>>> kinds of arguments and ranges. This makes them difficult to compare as 
>>>> they're not drop-in replacements for each other. Are there any 
>>>> particularly appropriate filter opcodes for this situation? Or any 
>>>> advice, techniques or tricks for doing this kind of thing would be 
>>>> greatly appreciated!
>>>>
>>>> Many thanks,
>>>>
>>>> Joe
>>>>
>>>
>>> -- 
>>>  Mark
>>>  _________________________________________
>>>  When you get lemons, you make lemonade.
>>>  When you get hardware, you make software.
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
>>> csound"
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
>> csound"
>>
>>
>
> -- 
>  Mark
>  _________________________________________
>  When you get lemons, you make lemonade.
>  When you get hardware, you make software.
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

victor wrote:
> Csound's reson is a 2-order all-pole resonator, with two poles that are
> complex-conjugate of each other at  Rexp(iw) and Rexp(-iw) (w = 
> 2*pi*fc/fs).

I just compared reson to mode, and they seem to give about the same 
results when you use the formula Q = frequency/bandwidth:

asig vco2 10, 137, 0
afilt1 reson asig, 200, 40
afilt2 mode asig, 200, 5
outs afilt1, afilt2*80

That means that the explanation on 
 is wrong, which is what I 
learned from originally. I actually wonder why reson and mode both exist.

> The filter in the link I sent is a common variation on the reson theme,
> with zeros placed at DC/Nyquist (so not all-pole, but poles+zeros).
> That corresponds to resonz in Csound. For more details you can
> follow the references in the csound manual page.

That makes sense, it just couldn't be simple resonance.

> I am not sure I follow what you are saying about bandpass having
> complex poles. IIR Filters that are 2-order or higher will exhibit
> complex-valued poles, regardless of the amplitude response shape
> (you can make them lowpass,bandpass,highpass).

I actually said that bandpass filters have real poles. This is because 
they are a lowpass filter followed by a highpass filter, which each have 
a real pole (the highpass filter also has a zero at the origin of the 
complex plane).

> ----- Original Message ----- From: "Mark Van Peteghem" 
> 
> To: 
> Sent: Saturday, May 02, 2009 6:50 AM
> Subject: [Csnd] Re: Re: Re: filters for physical modeling
>
>
>> That page is confusing to me, it talks about resonators but the graph 
>> is from a bandpass filter (0 gain at freq. 0, i.e. real poles). 
>> Resonance filters with complex poles have a gain that is not 0 when 
>> the frequency is zero, see . 
>> Maybe the word resonance is used differently in physics and digital 
>> music (I have a background in physics). And maybe one can use 
>> bandpass filters to make it sound good, although physics tells us to 
>> use resonance.
>>
>> I looked a little more at the reson opcode, and I'm confused here as 
>> well. The CSound manual says that reson is a second order _resonant_ 
>> filter (i.e. complex poles). But 
>>  (at about two thirds of 
>> the page) says it is a _band-pass_ filter; the graph clearly is that 
>> of a filter with real poles.
>>
>> In my first year at the university I had a course on vibrations and 
>> resonances, and it never mentioned band-pass filters, only resonance 
>> filters.
>>
>> victor wrote:
>>> Physical Models is not my area, but as far as I know, resonances
>>> are modelled by resonators, as discussed in
>>>
>>> http://www.music.mcgill.ca/~gary/307/week9/node4.html
>>>
>>> That's your reson filter alright.
>>>
>>> Victor
>>>
>>> ----- Original Message ----- From: "Mark Van Peteghem" 
>>> 
>>> To: 
>>> Sent: Friday, May 01, 2009 8:29 PM
>>> Subject: [Csnd] Re: filters for physical modeling
>>>
>>>
>>>> I guess band-pass filters may not be ideal for this. Resonances are 
>>>> better modeled with resonance filters, the only opcode I know that 
>>>> does it is mode (rezzy may also be good, but the manual says it is 
>>>> "created empirically", I don't know what it does). The opcode reson 
>>>> is a band-pass filter, so its name is actually misleading.
>>>>
>>>> Joseph Sanger wrote:
>>>>> Dear List,
>>>>>
>>>>> I'm trying to teach myself about Physical Modeling Synthesis at 
>>>>> the moment. In doing so I'm slowly getting to grips with filters.
>>>>>
>>>>> I've decided, as an exercise, to have a go at modeling a banjo. I 
>>>>> recorded the sound of tapping the banjo bridge with the strings 
>>>>> damped, and I identified some resonances of the banjo's head and 
>>>>> pot (using transforms in snd). I've started by trying various 
>>>>> band-pass filters to model the resonances, and playing the "pluck" 
>>>>> opcodes through them. It sounds OK... but not great!
>>>>>
>>>>> One problem I'm facing is that many of the filters take very 
>>>>> different kinds of arguments and ranges. This makes them difficult 
>>>>> to compare as they're not drop-in replacements for each other. Are 
>>>>> there any particularly appropriate filter opcodes for this 
>>>>> situation? Or any advice, techniques or tricks for doing this kind 
>>>>> of thing would be greatly appreciated!
>>>>>
>>>>> Many thanks,
>>>>>
>>>>> Joe
>>>>>
>>>>
>>>> -- 
>>>>  Mark
>>>>  _________________________________________
>>>>  When you get lemons, you make lemonade.
>>>>  When you get hardware, you make software.
>>>>
>>>>
>>>>
>>>> Send bugs reports to this list.
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>>> "unsubscribe csound"
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>> "unsubscribe csound"
>>>
>>>
>>
>> -- 
>>  Mark
>>  _________________________________________
>>  When you get lemons, you make lemonade.
>>  When you get hardware, you make software.
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>> "unsubscribe csound" 
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
> "unsubscribe csound"
>
>

-- 
  Mark
  _________________________________________
  When you get lemons, you make lemonade.
  When you get hardware, you make software.

Well, there are many types of bandpass filters. Reson is also a
bandpass filter, so I am not sure such generalisation is possible.
Also, I would have to check this, but putting 1st order 
lp & hp in cascade would make the resulting filter 2nd order and 
the resulting pair of poles would be most likely complex-valued. 
As I said, I need to check it up, because it's not something I ever 
looked into, but my intuition says that. First-order poles are of
course real-valued, on their own, but the resulting 2nd-order 
filter should have complex-valued poles. If I am wrong, I'd be
happy to be corrected.

> 
> I actually said that bandpass filters have real poles. This is because 
> they are a lowpass filter followed by a highpass filter, which each have 
> a real pole (the highpass filter also has a zero at the origin of the 
> complex plane).

Thank you, I'll try this.

Mark Van Peteghem wrote:
> Hi Joseph,
> 
> If mode isolates a frequency too much, you should lower the iq 
> parameter. I also noticed that you don't scale the signals created with 
> resonr, it may help to make it something like
> 
> aexcite    = (iscale1*afilt1+iscale2*afilt2+iscale3*afilt3+...)
> 
> 
> Mark
> 
> Joseph Sanger wrote:
>> Hi there Victor, Mark and List,
>>
>> Thanks for your replies, I'm at work now so I'll check the link later.
>>
>> Here is the csd of the banjo "so far". I'm fairly happy, but it's not 
>> finished yet! After experimenting with various filters I ended up with 
>> resonr - mode seemed to isolate the frequency too much.
>>
>> It seems that my problem has shifted to the string sound, rather than 
>> the resonating body. I haven't really started experimenting properly 
>> yet, but I want to model each of the five strings separately to allow 
>> for resonance and coupling between the strings (an important part of 
>> the banjo's sound!). I also want to model the banjo picks, and pick 
>> positions, so I need to really understand waveguide strings.
>>
>> Any advice would be welcome,
>>
>> Joe
>>
>> victor wrote:
>>> Physical Models is not my area, but as far as I know, resonances
>>> are modelled by resonators, as discussed in
>>>
>>> http://www.music.mcgill.ca/~gary/307/week9/node4.html
>>>
>>> That's your reson filter alright.
>>>
>>> Victor
>>>
>>> ----- Original Message ----- From: "Mark Van Peteghem" 
>>> 
>>> To: 
>>> Sent: Friday, May 01, 2009 8:29 PM
>>> Subject: [Csnd] Re: filters for physical modeling
>>>
>>>
>>>> I guess band-pass filters may not be ideal for this. Resonances are 
>>>> better modeled with resonance filters, the only opcode I know that 
>>>> does it is mode (rezzy may also be good, but the manual says it is 
>>>> "created empirically", I don't know what it does). The opcode reson 
>>>> is a band-pass filter, so its name is actually misleading.
>>>>
>>>> Joseph Sanger wrote:
>>>>> Dear List,
>>>>>
>>>>> I'm trying to teach myself about Physical Modeling Synthesis at the 
>>>>> moment. In doing so I'm slowly getting to grips with filters.
>>>>>
>>>>> I've decided, as an exercise, to have a go at modeling a banjo. I 
>>>>> recorded the sound of tapping the banjo bridge with the strings 
>>>>> damped, and I identified some resonances of the banjo's head and 
>>>>> pot (using transforms in snd). I've started by trying various 
>>>>> band-pass filters to model the resonances, and playing the "pluck" 
>>>>> opcodes through them. It sounds OK... but not great!
>>>>>
>>>>> One problem I'm facing is that many of the filters take very 
>>>>> different kinds of arguments and ranges. This makes them difficult 
>>>>> to compare as they're not drop-in replacements for each other. Are 
>>>>> there any particularly appropriate filter opcodes for this 
>>>>> situation? Or any advice, techniques or tricks for doing this kind 
>>>>> of thing would be greatly appreciated!
>>>>>
>>>>> Many thanks,
>>>>>
>>>>> Joe
>>>>>
>>>>
>>>> -- 
>>>>  Mark
>>>>  _________________________________________
>>>>  When you get lemons, you make lemonade.
>>>>  When you get hardware, you make software.
>>>>
>>>>
>>>>
>>>> Send bugs reports to this list.
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>>> "unsubscribe csound" 
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>> "unsubscribe csound"
>>>
>>>
>>
> 

-- 
Joseph Sanger

Blueberry Eikaiwa
www.bbeikaiwa.com
joe@bbeikaiwa.com

No, reson is not a bandpass filter. See 
 for an explanation of 
bandpass filters: they have a (approximately) flat frequency response in 
the frequency band that they let through. Resonance filters on the other 
hand have a resonance peak, their frequency response is not flat at all. 
Therefore bandpass filters are made from lowpass and highpass filters, 
AFAIK there is no other way.

If you put two or more filters in cascade, the resulting transfer 
function is simply the product of the transfer functions of each of the 
filters. Therefore every pole is preserved, unless it is annihilated by 
a zero, which is not the case here. And there can be no new poles. I 
must have studied over a thousand pages on transfer functions and 
related stuff as a student, so I'm quite sure about that.

Mark

victor wrote:
> Well, there are many types of bandpass filters. Reson is also a
> bandpass filter, so I am not sure such generalisation is possible.
> Also, I would have to check this, but putting 1st order lp & hp in 
> cascade would make the resulting filter 2nd order and the resulting 
> pair of poles would be most likely complex-valued. As I said, I need 
> to check it up, because it's not something I ever looked into, but my 
> intuition says that. First-order poles are of
> course real-valued, on their own, but the resulting 2nd-order filter 
> should have complex-valued poles. If I am wrong, I'd be
> happy to be corrected.
>
>>
>> I actually said that bandpass filters have real poles. This is 
>> because they are a lowpass filter followed by a highpass filter, 
>> which each have a real pole (the highpass filter also has a zero at 
>> the origin of the complex plane).
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
> "unsubscribe csound"
>
>

-- 
  Mark
  _________________________________________
  When you get lemons, you make lemonade.
  When you get hardware, you make software.

I'll have to disagree with you there. In my book, a bandpass filter is a 
filter whose
amplitude response passes signals at certain ranges and rejects signals
at others, regardless of their passband being flat or not.

Victor

----- Original Message ----- 
From: "Mark Van Peteghem" 
To: 
Sent: Sunday, May 03, 2009 9:41 AM
Subject: [Csnd] Re: Re: Re: Re: Re: Re: Re: filters for physical modeling


> No, reson is not a bandpass filter. See 
>  for an explanation of 
> bandpass filters: they have a (approximately) flat frequency response in 
> the frequency band that they let through. Resonance filters on the other 
> hand have a resonance peak, their frequency response is not flat at all. 
> Therefore bandpass filters are made from lowpass and highpass filters, 
> AFAIK there is no other way.
>
> If you put two or more filters in cascade, the resulting transfer function 
> is simply the product of the transfer functions of each of the filters. 
> Therefore every pole is preserved, unless it is annihilated by a zero, 
> which is not the case here. And there can be no new poles. I must have 
> studied over a thousand pages on transfer functions and related stuff as a 
> student, so I'm quite sure about that.
>
> Mark
>
> victor wrote:
>> Well, there are many types of bandpass filters. Reson is also a
>> bandpass filter, so I am not sure such generalisation is possible.
>> Also, I would have to check this, but putting 1st order lp & hp in 
>> cascade would make the resulting filter 2nd order and the resulting pair 
>> of poles would be most likely complex-valued. As I said, I need to check 
>> it up, because it's not something I ever looked into, but my intuition 
>> says that. First-order poles are of
>> course real-valued, on their own, but the resulting 2nd-order filter 
>> should have complex-valued poles. If I am wrong, I'd be
>> happy to be corrected.
>>
>>>
>>> I actually said that bandpass filters have real poles. This is because 
>>> they are a lowpass filter followed by a highpass filter, which each have 
>>> a real pole (the highpass filter also has a zero at the origin of the 
>>> complex plane).
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
>> csound"
>>
>>
>
> -- 
>  Mark
>  _________________________________________
>  When you get lemons, you make lemonade.
>  When you get hardware, you make software.
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound"