phonTools (version 0.2-2.2)

freqresponse: Frequency Response

Description

Find the frequency response of a digital filter

Usage

freqresponse (b, a, fs = 0, add = FALSE, show = TRUE,
              steps = 1000,...)

Value

A dataframe with two columns (frequency and response) that can be used to redraw the frequency response if required. The 'response' value corresponds to dB. magnitude below peak.

Arguments

b

The moving-average (MA), numerator coefficients of the filter.

a

The autoregressive (AR), denominator coefficients of the filter. Please note that the leading 1 at a[0] is not assumed.

fs

The sampling frequency of the sound. If this is not given calculations are presented as if fs = 1.

add

If TRUE, the frequency response plot is added to an existing plot.

show

If TRUE, the frequency response of the estimated filter is plotted.

steps

The number of steps between zero and the Nyquist frequency.

...

Additional arguments are passed to internal plotting functions.

Author

Santiago Barreda <[e-mail]>

Details

This function plots (and optionally returns) the frequency response for the digital filter whose transfer function is determined by the numerator and denominator filter coefficients given in b and a.

References

Lyons, R. G. (2004). Understanding Digital Signal Processing (2nd ed.). Prentice Hall.

Examples

## make a synthetic vowel with a known set of formant frequencies
#sound = vowelsynth (ffs = c(500,1500,2500,3500,4500), 
#fbw = c(30, 80, 150, 200, 220),f0 = 100, dur = 100)

#plot (sound)

## let the LPC function estimate the filter used to generate the vowel
#coeffs = lpc (sound, show = FALSE)

## compare frequency response of estimated filter to vowel spectrum
#spectralslice (sound, col = 4, preemphasisf = 50)
#freqresponse (1, coeffs, add = TRUE, fs = 10000)

## generate a sinc function
#filt = sinc (seq (-15,15,1/2), normalized = TRUE)
## treat it as a low-pass FIR filter and inspect its frequency response
#freqresponse (filt, 1)