|Filter: the canonical-form n-pole/m-zero filter class. More...|
|Notch Filter with poles at +-z and complex conjugate zeros at +-omega. More...|
|Moog VCF class. More...|
|< we use the STL vector, map, and string classes |
|Declare the pointer to freq/bw buffers (if used) and current scale/offset values. |
|Load the freq/bw-related values at the start. |
|#define FILTER_MAX_COEFFICIENTS (16)|
Filters.h -- CSL filter classes.
The base Filter class can perform the generic filter equation based upon a set of feedback and feedforward coefficients. Subclasses of Filter implement the setupCoeffs method to build these coefficients according to different algorithms. The subclasses here mostly inherit from the FrequencyAmount controllable, which specifies a center frequency and an 'amount' which may variously be resonance, radius, bandwidth etc according to the algorithm.
OK so for example Butter class has a CenterFrequency port and a Bandwidth port, but pull_controls must be calling Controllables to do the pullInput()
Or, a generic filter that can take a multichannel UGen for each of aCoeffs and bCoeffs (i.e. multichannel ports), and other filter classes that wrap this and have Frequency and Bandwidth ports etc. This reduces the calls to setupCoeffs, because they'd actually be part of the nextBuffer instead
The reason is that the Scalable type approach & macros can't be extended to filter otherwise. ALSO, there are no similar macros for Effect; how should this work?
OK, the way he had it working here is filtering in place, i.e. no internal buffer, but if I inherit from Effect, I do have an internal buffer; this means the first thing to do is memcopy the input to the output, then pounce on that; or do the in-place stuff in the Effect port and finally copy to output, say with scale & offset performed there.
See the copyright notice and acknowledgment of authors in the file COPYRIGHT
|#define BW_LOW_PASS 0|
|#define BW_HIGH_PASS 1|
|#define BW_BAND_PASS 2|