naturerecordists
[Top] [All Lists]

Re: frequency and spectrograms

Subject: Re: frequency and spectrograms
From: Walter Knapp <>
Date: Fri, 15 Mar 2002 18:14:42 -0500
olarte1 wrote:
> 
> Hello,
> 
> Until now I've been digitizing my analog cassette recordings at 16
> bit/48khz. Now am told that it is an overkill and I should sample at
> 44.1Khz, others tell me I will always get better spectrograms at
> 48KHz than at 44.1Khz.  I record with type II cassetes which means a
> max. of 14khz. If I go down, and digitize at 44.1 will I leave
> something behind on those cassettes?

As a general rule you get frequencies up to half the sampling frequency.
So, 44.1 is good for 22khz. In actual practice that upper limit is only
done well for the rather rare sample that's exactly in phase with the
original sound. I usually set a upper limit around 15 - 16 khz for what
I trust with digital sound. So, either will easily handle your cassettes.

You might get better spectrograms if the software you are using is
optimized to give better ones at 48khz and not for 44khz. But, I've
found that it's rarely a problem, that other things make far more of a
problem. The software I use does equally well with either. I routinely
use 44khz and do sonograms that work out just fine. You can look at some
on my website:
http://wwknapp.home.mindspring.com/GAFrog.Toad.html

Note these were done several years ago. I've another piece of software
I'm beginning to like as it can do log scale sonograms, that's SparkXL
(it's a mac program). In either my older program (Soundhack) or this new
one there is no difference in what I see from either sample rate, in
fact from even much lower sample rates. With Soundhack the only way you
could set the frequency scale is by changing the sample rate, it's scale
covers half of whatever sample rate you use. So you will find that most
of the sonograms on my pages were done from downsampled files to 11khz
or 22khz sample rates.  SparkXL does not have this limit, and also has
some choices as to resolution. So, it may show a slight bit more detail.

The reason for using 16bit 44.1khz is that it's the same as audio CD's,
so no conversion when you write those. Computer software for general
audio use tends to be optimized for 44khz as a result. 48khz is
associated with DAT, but is not really any particular improvement. In
theory it gives you a extra khz or so, but unless your source has those
it's not important. And it's a conversion to avoid.

Walt



________________________________________________________________________
________________________________________________________________________

<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU