naturerecordists
[Top] [All Lists]

RE: Difficult Speech

Subject: RE: Difficult Speech
From: Michael Dalton <>
Date: Thu, 14 Apr 2005 08:29:13 -0700 (PDT)
Thanks to those recordists who responded to my
questions.

In the recent discussion, several commented about my
bird's physical situation. Here are additional
details: the macaw (a pet and research subject) is
located outside close to a concrete wall of an
"L-shaped" part of the house. Arielle (the bird) stays
outside all day on a gym located on the porch. She
speaks from a location that faces a city street, only
about 60 feet away, that contributes the vehicular
noise present in my recordings. There does not appear
to be a practical way to treat the area with sound
absorbing materials, nor is there a better location
for the bird. The bird has been put outside for more
than 12 years; that is her routine, so it is unlikely
that I will find a better spot to record her speech
(the subject speaks voluntarily).

I appreciate comments made by several recordists.
KennJava and Dan Dugan's dialogue about inverting the
sound in one channel and using the mixer to subtract
off the noise was of interest. I tried that experiment
with a pair of microphones in the house to try to
eliminate room effects, but there were no significant
results. I gave up on that idea. Additionally, the
noise is complicated by factors that I am just
beginning to comprehend. There is Doppler shift in the
spectrum caused by the movement of the cars along the
road. There are reflections from the building,
diffracted sound from an intervening fence, and many
unexpected "virtual" sources.

The computer treatment of the files seems to be a good
prospect for making the recordings more
understandable. I am looking into some noise reduction
software. (I browsed the Pristine Sounds site
suggested by Dan Dugan.)

People encourage me to make better recordings, and I
have now switched to noisy recordings stored on CD. If
I were able to make perfect recordings with no noise,
I would likely still have problems. I am finding that
people aren't very good at listening to mechanically
reproduced sounds, especially short excerpts, although
listeners aren't very good at the task live either.
Interested parties might want to read a newspaper
reporter's impressions in an article about Arielle,
"From the beak of a Macaw," at the following location.
http://www.sptimes.com/2003/05/12/Northpinellas/From_the_beak_of_a_ma.shtml


The previous topic brings me to the next point. While
there is sometimes heavy noise on the program, I can
still make out the words. (Sure, I have an advantage
as the owner, but most listeners perceive only about
4-8% of what the bird says.) I think that someone else
who had amateur radio experience (as I did as a kid)
or "a nature recordist" might be better able to
understand speech signals embedded in noise. I know
there must be an experienced listener who can bring
his skill to bear upon the information. Some people
understand some of the unusual words spoken by Arielle
in isolated, random, instances. This indicates to me
that they are hearing the sounds, but they process the
words inconsistently. For the average listener, it
appears that the sound is disconnected from the
corresponding words in the listener's brain thereby
preventing interpretation of the messages. (I'm going
to send a couple of files to individuals who expressed
interest in hearing a sample. However, I cannot do so
immediately because I am in the process of purchasing
hardware and software to convert selected samples from
my audiocassette.)

Further thoughts?

Mike Dalton
Florida





=09=09
__________________________________
Do you Yahoo!?
Yahoo! Small Business - Try our new resources site!
http://smallbusiness.yahoo.com/resources/


________________________________________________________________________
________________________________________________________________________

<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