TableInterpFract32

Overview

Fract32 Table driven interpolation module. Supports linear and spline modes

Discussion

This module performs table lookup using defined X and Y points. The module supports linear and pchip interpolation. Internally the interpolation is performed using a piece wise polynomial approximation. With linear interpolation a first order polynomial is used; for pchip a 4th order polynomial is used. PChip is similar to a spline interpolation but it avoids the overshoot and undershoot that plagues splines.

The input to the module serves as the X value in the lookup and the output equals the Y value derived from doing the interpolation. The module is configured by setting the internal matrix .XY. The first row of .XY represents X values and the second row Y values.

At instantiation time you specify the total number of points (MAXPOINTS) in the table. MAXPOINTS determines the amount of memory allocated to the table. Then at run time, you can change the number of points actively used by the table (.numPoints). .numPoints must be less than or equal to MAXPOINTS.

The variable .order specifes either linear (=2) or cubic (=4) interpolation.

The module also provides a custom inspector for easy configuration from the Audio Weaver Designer GUI. The GUI translates the XY table into a piece wise polynomial segments. The matrix .polyCoeffs contains 4 x (MAXPOINT-1) values. Each column represents the coefficients for each polynomial. Row 1 is the X^3 coefficent, row 2 the X^2 coefficient, row 3 the X coefficient and row 4 the constant term.

If the input x falls outside of the range of values in the XY table then the input is clipped to the allowable range.

The X values in the table have to monotonically increasing. That is, x0 < x1 < x2 < ... < xN. If this condition is not obeyed, then the module's prebuild function will generate an error.

Module Pack

Standard

ClassID

classID = 211

Type Definition

typedef struct _ModuleTableInterpFract32
{
ModuleInstanceDescriptor instance;            // Common Audio Weaver module instance structure
INT32 maxPoints;                              // Maximum number of values in the lookup table. The total table size is [maxPoints 2].
INT32 numPoints;                              // Current number of interpolation values in use.
INT32 outRange;                               // Indicates if coefficients are out of fractional range.
INT32 alerts;                                 // Indicates if coefficients are out of fractional range.
INT32 order;                                  // Order of the interpolation. This can be either 2 (for linear) or 4 (for pchip).
fract32* XY;                                  // Lookup table. The first row is the X values and the second row is the Y values.
fract32* polyCoeffs;                          // Interpolation coefficients returned by the grid control.
} ModuleTableInterpFract32Class;

Variables

Properties

Name	Type	Usage	Default Value	Range
maxPoints	int	const	8	4:1:1000
numPoints	int	parameter	7	4:1:8
outRange	int	derived	0	0:1:1
alerts	int	derived	0	0:1:1
order	int	parameter	2	2:2:4
XY	fract32*	parameter	[2 x 8]	Unrestricted
polyCoeffs	fract32*	state	[4 x 7]	Unrestricted

Pins

Input Pins


Name	in
Description	audio input
Data type	fract32
Channel range	Unrestricted
Block size range	Unrestricted
Sample rate range	Unrestricted
Complex support	Real

Output Pins


Name	out
Description	audio output
Data Type	fract32

Matlab Usage

File Name: table_interp_fract32_module.m 
 M=table_interp_fract32_module(NAME, MAXPOINTS) 
 This Audio Weaver module performs interpolation using a lookup table together 
 with a configurable interpolation order. The table contains (X,Y) value pairs 
 and can be unevenly spaced. 

 Arguments: 
    NAME - name of the module. 
    MAXPOINTS - maximum number of points allocated to the lookup table. This 
         is set at design time and has a minimum value of 4. At run 
         time, you can change the number of values in the lookup table 
         from 4 to MAXPOINTS. 

 The module can be configured to perform linear or pchip interpolation.