Filter Class Reference

Algorithm to apply one or several IIR filters given the Real value coefficients. More...

#include <Filter.h>

List of all members.

Public Member Functions
	Filter (int channelCount=1)
void	process (const MatrixXR &samples, MatrixXR *filtered)
void	a (MatrixXR *a) const
void	setA (const MatrixXR &a, bool callSetup=true)
void	b (MatrixXR *b) const
void	setB (const MatrixXR &b, bool callSetup=true)
int	channelCount () const
void	setChannelCount (int count, bool callSetup=true)
int	length () const

---

Detailed Description

Algorithm to apply one or several IIR filters given the Real value coefficients.

This class represents an object to apply one or several IIR filters. The coefficients must be manually set by the user. To create and use some special parametrized filters such as Low Pass, High Pass, Band Pass and Band Stop filters refer to IIRFilter.

For Complex coefficient filters, refer to FilterComplex.

This filter implementation allows single and multiple channel filtering. The number of channels is specified using the setChannelCount(). The a and b coefficients of the filter are specified by two matrices of Real values. The rows of the matrix are the time indices of the filter and the columns (if more than one) are the channels.

Three different situations are possible, depending on the number of columns in the coefficients matrices and in the input matrix:

• if the number of columns of the coefficients matrices are equal to the number columns in the input matrix, then each column of the input matrix is filtered by a column of the coefficients matrix. This is the situation when trying to filter differently all the channels of a multi-channel signal.
• if the coefficients matrix has one single column and the input matrix has multiple columns then each column of the input matrix is filtered by the single column of the coefficients matrices. This is the situation when trying to filter equally all the channels of a multi-channel signal.
• if the coefficients matrices have multiple columns each and the input matrix has multiple columns then the column of the input matrix is filtered by each of the columns of the coefficients matrices. This is the situation when applying a filterbank to a single channel signal.

Note that in all cases the number of columns in a and b coefficients matrices must be the same. Note that the channel count determines the number of output channels in any situation and is therefore must be equal to the maximum number of channels between input and coefficient matrices.

Author:

Ricard Marxer

See also:

Filter

---

Constructor & Destructor Documentation

Filter

(

int

channelCount = 1

)

Constructs a band pass filter object with the given channelCount, b, and a coefficients given.

References setA(), setB(), and setChannelCount().

---

Member Function Documentation

void process	(	const MatrixXR &	samples,
		MatrixXR *	filtered
	)

Performs a filtering of each of the columns of samples. Puts the resulting filtered in the columns of filtered.

Parameters:

	samples	matrix of Real values. A column represents a channel and a row represents a time index.
	filtered	pointer to a matrix of Real values for the output. The matrix should have the same number of rows and columns as samples.

Note that if the output matrix is not of the required size it will be resized, reallocating a new memory space if necessary.

See also:

IIRFilter::process

Referenced by LPC::process(), and BandFilter::process().

void a

(

MatrixXR *

a

)

const

Returns the a coefficients of the filter

Both the b and a coefficients matrices are normalized by the first row of the a coefficients matrix.

Note that if the first row of the a coefficients matrix has elements to zero, some of the filtered samples will result in NaN.

Note that the number of columns in a and b must be the same, and that it must be equal to 1 or Filter::channelCount.

By default it is a single element matrix of value 1.0.

See also:

setA(), b(), setB()

Referenced by BandFilter::a().

void setA	(	const MatrixXR &	a,
		bool	callSetup = true
	)

Specifies the a coefficients of the filter

Both the b and a coefficients matrices are normalized by the first row of the a coefficients matrix.

Note that if the first row of the a coefficients matrix has elements to zero, some of the filtered samples will result in NaN.

Note that the number of columns in a and b must be the same, and that it must be equal to 1 or Filter::channelCount.

See also:

a(), b(), setB()

Referenced by Filter().

void b

(

MatrixXR *

b

)

const

Returns the b coefficients of the filter

Both the b and a coefficients matrices are normalized by the first row of the a coefficients matrix.

Note that if the first row of the a coefficients matrix has elements to zero, some of the filtered samples will result in NaN.

Note that the number of columns in a and b must be the same, and that it must be equal to 1 or the channel count.

By default it is a single element matrix of value 1.0.

See also:

setB(), a(), setA()

Referenced by BandFilter::b().

void setB	(	const MatrixXR &	b,
		bool	callSetup = true
	)

Specifies the b coefficients of the filter

Both the b and a coefficients matrices are normalized by the first row of the a coefficients matrix.

Note that if the first row of the a coefficients matrix has elements to zero, some of the filtered samples will result in NaN.

Note that the number of columns in a and b must be the same, and that it must be equal to 1 or the channel count.

See also:

b(), a(), setA()

Referenced by Filter().

int channelCount

(

)

const

Returns the number of output channles of the filter

Note that the number of channels must be equal to the number of columns in a and b or to the number of columns in the input matrix.

By default it is 1.

void setChannelCount	(	int	count,
		bool	callSetup = true
	)

Specifies the number of output channles of the filter

Note that the number of channels must be equal to the number of columns in a and b or to the number of columns in the input matrix.

Referenced by Filter().

int length

(

)

const

Returns the length of the filter, which is the maximum number of rows between the a and b coefficients matrices.

---

The documentation for this class was generated from the following files:

• Filter.h
• Filter.cpp