remezFIRFilter

import signals.filters.tfilter-1;

Design filter



filter=remezFIRFilter(0, 2 kHz, [
		{
			gain: 1, // passband
			ripple: 5dB,
			start: 0Hz,
			stop: 400Hz
		},
				   
		{
			gain: 0, // stopband
			atten: -40dB,
			start: 500Hz,
			stop: 1000Hz
		}
	]);
filter=remezFIRFilter(0, 2 kHz, [
		{
			gain: 0, // stopband
			atten: -40dB,
			start: 0Hz,
			stop: 400Hz
		},
				   
		{
			gain: 1, // passband
			ripple: 5dB,
			start: 500Hz,
			stop: 1000Hz
		}
	]);
filter=remezFIRFilter(0, 2 kHz, [
		{
			gain: 0, // stopband
			atten: -40dB,
			start: 0Hz,
			stop: 200Hz
		},
				   
		{
			gain: 1, // passband
			ripple: 5dB,
			start: 300Hz,
			stop: 500Hz
		},

		{
			gain: 0, // stopband
			atten: -40dB,
			start: 600Hz,
			stop: 1000Hz
		},
	]);
filter=remezFIRFilter(0, 2 kHz, [
		{
			gain: 1, // passband
			ripple: 5dB,
			start: 0Hz,
			stop: 200Hz
		},
				   
		{
			gain: 0, // stopband
			atten: -40dB,
			start: 300Hz,
			stop: 500Hz
		},

		{
			gain: 1, // passband
			ripple: 5dB,
			start: 600Hz,
			stop: 1000Hz
		},
	]);
filter.plot()

If num_taps is zero or null, algorithm will try to find the minimal suitable tap number.

Bands are specified by the ibands argument. It is a list of frequency specifications. Every element is an object with the following fields:

gain: desired gain. Usually 1 for passband, 0 for stopband.
ripple: use only for passband: the passband ripple.
atten: use only for stopband: stopband attenuation.
start: band start frequency.
end: band end frequency.

You should leave a gap (transition band) between the specified bands, otherwise the filter will not be realizable and the algorithm will not converge.

You can use arbitary number of bands.

It returns a signals.containers-1/TransferFunction structure.

import signals.filters.tfilter-1;

Get number of taps


filter.num.length

import signals.filters.tfilter-1;

Get taps


filter.num

Based on TFilter (http://t-filter.appspot.com).

For further details on FIR filters, see http://en.wikipedia.org/wiki/Finite_impulse_response.

Read more about the Remez algorithm at http://en.wikipedia.org/wiki/Remez_algorithm.

numTaps	the number of taps wanted. If 0 or null, the smallest number of taps will be used.
sampFreq	the sample frequency.
bands	the band specification array.
return value	linear system container (TransferFunction).

remezFIRFilter(numTaps, sampFreq, bands)