Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wuarchive!uunet!mcsun!cernvax!chx400!bernina!hanke
From: hanke@nessie.cs.id.ethz.ch (Norbert Hanke)
Newsgroups: comp.dsp
Subject: Re: 48k to 44.1k sample rate conversion
Message-ID: <1991May17.075009.10410@bernina.ethz.ch>
Date: 17 May 91 07:50:09 GMT
References: <1991May13.173129.18295@bernina.ethz.ch> <1991May15.105543.12165@bernina.ethz.ch> <1991May16.052920.21060@netcom.COM>
Sender: news@bernina.ethz.ch (USENET News System)
Organization: Swiss Federal Institute of Technology (ETH), Zurich, CH
Lines: 23
Nntp-Posting-Host: nessie

In article <1991May16.052920.21060@netcom.COM> mfvargo@netcom.COM (Michael Vargo) writes:
(some stuff deleted)
>1. If there really is significant energy in the 22,500 to 24,000Hz band,
>then a low pass filter will be required to prevent aliasing on the
>downsample.  Maybe this signal is dog whistles so something will be needed.
>It seems a simple FIR filter could be applied to the 48kHz sampled data to
>wipe this out.  I have an Excel spreadsheet to calculate coefs if your

The ~10000-Tap FIR filter I described earlier does exactly that. It has
to be designed as a FIR filter for a sampling frequency of 7.056 MHz with
the following properties: passband 0..20 kHz, stopband 24.1 kHz..fs/2
(or 22.05 kHz..fs/2 if you don't want to have alias products in the range
20..22.05 kHz).

>Does the act of interpolation actually do some sort of low pass filtering?
>I think not but maybe it could be looked at as some sort of averaging?

An FIR filter does nothing more than a weighted interpolation between samples.
Simple interpolation is FIR filtering with all filter coefficients having the
same value, which results in a sin(x)/x - shaped frequency response.

Norbert Hanke
ETH Zurich, Switzerland