naturerecordists
[Top] [All Lists]

Re: Sample Rate Conversion

Subject: Re: Sample Rate Conversion
From: Walter Knapp <>
Date: Sat, 15 May 2004 00:59:09 -0400
From: Vicki Powys <>
> 
> Walt wrote:
> 
> 
> 
>>> On the basis of your statements that you believe yourself to have very
>>> good high frequency hearing, I would think you probably are judging
>>> these samples on the basis of the high frequency band. If you have a cut
>>> filter available, try filtering out everything above 10khz or so and
>>> listen to the lower stuff. That was something else I did with these
>>> samples was to cut out various bands so as to listen to just each part.
> 
> 
> 
> Well I've tried it the other way around, just to see what I am basing my
> judgement on, in comparing these insect sounds.  And yes I am definitely
> judging the insect sounds from the high frequency band.  I've just tried
> using the Telephone Filter (i.e. brick wall filter) in Waves Q10 plug-in in
> Peak LE.  I've used this as a low-cut filter just to see how high I can hear
> this insect sound.  I can run the low cut filter right off the dial at the
> high end without adjusting the volume, and there is still some audible sound
> right up in the 20 kHz end.  It is hard to tell just what the reading really
> is, but this is how I'm measuring it:  In Q10 I have buttons 1-2-3-4-5 all
> ON and 6-7-8-9-10 all OFF.  I am adjusting the frequency by shift-clicking
> 1-2-3-4-5 in the frequency column, to drag the low cut further to the right.
> Taking my reading from 2-3-4 (which are all the same), I am hearing a clear
> sound and the same loudness up to 14 kHz.  At 14 kHz the loudness is very
> slightly less, then tapers off as I go upwards but I can still just hear
> some sound when 2-3-4 read 20 kHz.  (1 & 5 have a different reading to
> this.)

As i noted I can hear the full range, but as the frequency increases I 
have to increase the volume.

> I had a look at the various insect files in Audiograph and you are quite
> right, there is a very faint band of sound between 4.5 & 5 kHz.  The main
> band shows very clearly at 10-16 kHz.

There is a even fainter bit of sound below 160hz, but you will need to 
be in log scale to see it.

Note I think the 5khz sound belongs and is part of the noise made by the 
insect in generating the higher frequencies.

> I tried converting the 48k file first to 65k then to 44k (in Peak LE) but it
> sounded identical to a file converted direct to 44k.  In 'Save As' I only
> have the option of 16 bit or 8 bit, so that wasn't any help.

In peak's conversion try looking for faint sound below the band that 
belongs, between there and about 2khz. That's it's main error, that 
faint sound. If you filter out all the frequencies above about 7khz, you 
can probably hear the differences in that.

Clearly the OSX Peak I use has a bunch of added stuff as people move to 
higher sampling rates and bit depths. It can also be that my hardware is 
capable of more than the iMac, as what Peak offers is partially based on 
the hardware it finds.

> If nothing else, this exercise is a good way of learning more about using
> sound software!  Thanks for your tips Walt.

That's a lot of how I come up with software tricks, just play with 
things and see what I get. I approach problems like this resampling one 
the same way. I'm not so much worried about fixing blame, but how I 
might can get around the problem.

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