Exp() is handled by a function table, not a function call. It's just
waveshaping+ring modulation. Waveshaping used to have a bad name
because of  polynomial transfer functions which are almost useless for
dynamic spectra. But not this one (and there are others).

The DCBlocker is OK, not really costly (no need to be linear-phase)
As fast or faster than some popular methods (Blep etc), no
divisions, much cleaner than DPW
.
It's actually pretty fast, I have a Csound reference
implementation. It's also being added to a commercial software synth by
the people we wrote it for.

Victor

----- Original Message ----- 
From: "Richard Dobson" 
To: 
Sent: Monday, November 24, 2008 9:20 PM
Subject: [Csnd] Re: Re: Re: UDO question..


> Indeed - wish I had been there!  I have wondered how fast it is (including 
> DC blocking filter - linear-phase one?), given it requires an exp call per 
> sample, and users typically want a synth supporting ~lots~ of voices.
>
> Richard Dobson
>
>
>
> victor wrote:
>> In fact, there are ways of generating nearly band-limited signals
>> almost that cheaply, which can be smoothly dynamically changed.
>> We had a paper on one such method at the last DAFx and another
>> under review discussing a more generalised solution. Also currently
>> working on a related idea for resonant cases.
>>
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

victor wrote:
> 
...
> .
> It's actually pretty fast, I have a Csound reference
> implementation. It's also being added to a commercial software synth by
> the people we wrote it for.
> 

Sounds good. Will there be any obstacles to a published Csound opcode?


Richard Dobson