Release 11 Enhancements (Release 11 New Features)

Release 11 New Features

Signal Processing Toolbox 4.2

Version 4.2 of the Signal Processing Toolbox delivers a number of improvements and enhancements, described below.

Also see the Signal Processing Toolbox readme file for a summary of the new additions. To view the readme file, type

info signal

New Functions

Name
Purpose

ac2poly
Autocorrelation sequence to prediction polynomial conversion

ac2rc
Autocorrelation sequence to reflection coefficients conversion

arburg
AR parametric modeling via Burg's method

arcov
AR parametric modeling via the covariance method

armcov
AR parametric modeling via the modified covariance method

aryule
AR parametric modeling via the Yule-Walker method

buffer
Buffer a signal vector into a matrix of data frames

pcov
Power Spectrum estimate via the covariance method

pmcov
Power Spectrum estimate via the modified covariance method

poly2ac
Prediction polynomial to autocorrelation sequence conversion

pwelch
Power Spectrum estimate via Welch's modified periodogram method

rc2ac
Reflection coefficients to autocorrelation sequence conversion

rlevinson
Reverse Levinson-Durbin recursion

sgolay
Design a Savitzky-Golay smoothing filter

sgolayfilt
Filter a signal with a Savitzky-Golay smoothing filter

sosfilt
Filter a signal using second-order sections (biquad)

tf2sos
Transfer function to second-order sections conversion

Name	Purpose
`ac2poly`	Autocorrelation sequence to prediction polynomial conversion
`ac2rc`	Autocorrelation sequence to reflection coefficients conversion
`arburg`	AR parametric modeling via Burg's method
`arcov`	AR parametric modeling via the covariance method
`armcov`	AR parametric modeling via the modified covariance method
`aryule`	AR parametric modeling via the Yule-Walker method
`buffer`	Buffer a signal vector into a matrix of data frames
`pcov`	Power Spectrum estimate via the covariance method
`pmcov`	Power Spectrum estimate via the modified covariance method
`poly2ac`	Prediction polynomial to autocorrelation sequence conversion
`pwelch`	Power Spectrum estimate via Welch's modified periodogram method
`rc2ac`	Reflection coefficients to autocorrelation sequence conversion
`rlevinson`	Reverse Levinson-Durbin recursion
`sgolay`	Design a Savitzky-Golay smoothing filter
`sgolayfilt`	Filter a signal with a Savitzky-Golay smoothing filter
`sosfilt`	Filter a signal using second-order sections (biquad)
`tf2sos`	Transfer function to second-order sections conversion

New Demos

Name
Purpose

sgolaydemo
Demonstrates Savitzky-Golay filtering

Name	Purpose
`sgolaydemo`	Demonstrates Savitzky-Golay filtering

Enhanced Functions

firrcos

The firrcos function now:

Accepts either a bandwidth or a roll-off factor
Designs either a normal or square root raised cosine filter
Accepts an arbitrary variable delay for the impulse response
Accepts a window parameter for the filter design

pburg, pmtm, pmusic, pyulear

These functions have changed in a way that may affect your results.

When no sampling frequency (Fs) is specified, these functions return the PSD estimate, Pxx(

), as a function of normalized angular frequency,

in rads/sample. If Fs is specified, the functions return the PSD estimate,
Pxx(f)/Fs, as a function of linear frequency, f, in Hz. Fs defaults to 1 Hz. Note that the new functions pcov, pmcov, and pwelch also adhere to this specification.

poly2rc

Returns the zero-lag autocorrelation when called with the optional second input argument, the final prediction error.

rc2poly

The rc2poly function has changed in ways that may affect your results:

Returns a column vector rather than a row vector
Returns the final prediction error when called with the optional second input argument, the zero lag autocorrelation

sos2ss, sos2tf, sos2zp

The sos2ss, sos2tf, and sos2zp functions now accept an optional second input argument, the gain returned by the functions that convert to second-order-sections (SOS) form (ss2sos, tf2sos, and zp2sos).

ss2sos, zp2sos

The ss2sos and zp2sos functions have changed in ways that may affect your results. These functions provide an additional output argument corresponding to the gain of the second-order-sections structure and also accept an additional input argument that specifies the desired scaling of the structure. Scaling choices are:

-norm, 2-norm, and none.

detrend Now Part of MATLAB Language

The detrend function now ships in the toolbox/matlab/datafun directory as part of the standard MATLAB language.

Interactive Tool Enhancements

The following tools have changed in ways that may affect your results.

SPTool loads a default session upon startup.
Signal Browser offers printing with preview.
Filter Designer provides a Pole/Zero Editor to supplement the existing design methods.
Spectrum Viewer provides new covariance and modified covariance spectral estimation methods, as well as printing with preview. Additionally,
- The maximum entropy method (MEM) has been removed. Use the Yule-Walker AR method instead.
- Welch's method now uses the pwelch function instead of psd, and therefore no longer offers the scaling or detrending options. (pwelch internally scales the PSD magnitude by 1/Fs, and does not detrend the original signal.)
- The Burg and Yule-Walker AR methods now scale the PSD magnitude by 1/Fs
- The option to specify an autocorrelation matrix as input to the Yule-Walker AR method has been removed

[ Previous | Next ]