Thank you Brian - for those without access to the document, the following is the relevant segment from the Discussion:
The subjects could discriminate the stimuli with and without ultrasounds only when all components were mixed and presented through the same loudspeaker. When as many as 6 loudspeakers were used in order to prevent intermodulation, no one could discriminate the stimuli. No significant difference was found between monaural and stereophonic conditions. It was indicated from the results that the non-linear interaction of ultrasounds in the air or in the auditory system was, if any, not so much as that in the average loudspeakers as far as the level of the signal did not exceed 80 dB SPL. It was shown that under conditions in which experimental artifacts had been adequately eliminated, ultrasounds would be extremely difficult to be perceived. They may have little influence on the sound image and its localization
Basically in the single-speaker case (response of the speaker claimed to 50kHz) measurable intermodulation distortion was documented, leading to the ability to detect the ultrasound harmonics (since these artifacts started in the sub-20kHz region).
When intermodulation distortion was avoided (using individual speakers to deliver each harmonic beyond 20kHz) no detectability for the ultrasound components was seen in the test subjects.
And yes, the issue below with the Matlab below is aliasing - a rather fundamental issue when attempting to synthesize waveforms.
- Neil
On Feb 4, 2011, at 11:18 AM, Brian C. J. Moore wrote:
Deall All,
In connection with this thread, I suggest taking a look at:
Ashihara, K. and Kiryu, S. (2000).
Influence of expanded frequency band of signals on non-linear
characteristics of loudspeakers, J. Acoust. Soc. Jap. (J)
56, 549-555.
Ashihara, K. and Kiryu, S. (2003). Audibility of components above 22 kHz
in a complex tone, Acustica - acta acustica 89,
540-546.
Brian Moore
Just to show that it is not a
simple question of loudspeaker quality but also of recording quality I
send you a Matlab script producing one square wave "sampled" at
48000 and another at 192000 Hz. The difference is audible through
any loudspeaker.
clear all
sf1 = 48000;
sf2 = 192000;
dt1 = 1/sf1;
dt2 = 1/sf2;
du = 1;
f0 = 5000;
t1 = 0:dt1:du;
t2 = 0:dt2:du;
s1 = square(2*pi*f0*t1);
s2 = square(2*pi*f0*t2);
sound(s1, sf1)
wavwrite(s1, sf1, 's48000')
pause(1)
sound(s2, sf2)
wavwrite(s2, sf2, 's192000')
Best,
Dik
> -----Original Message-----
> From: AUDITORY - Research in Auditory Perception
>
[
mailto:AUDITORY@xxxxxxxxxxxxxxx] On Behalf Of Joachim Thiemann
> Sent: vrijdag 4 februari 2011 15:37
> To: AUDITORY@xxxxxxxxxxxxxxx
> Subject: Re: [AUDITORY] High-frequency hearing in humans
>
> On Thu, Feb 3, 2011 at 23:08, Kevin Austin
<kevin.austin@xxxxxxxxxxxx>
> wrote:
> > A colleague of mine has been working on clicks in an
electroacoustics
> aural perception course. He discovered that a 48kHz sampling rate
was
> "too crude", and that working at 96kHz (or higher), the
differences
> between clicks over 8kHz were noticeable. This is not quite
what
> sampling theory seems to say. At 44.1kHz, a single sample click
> represents 22kHz, and a two sample click represents 11kHz. How does
one
> represent a 16kHz click with a 44.1kHz sampling rate?
>
> I think one has to be careful about the actual D/A hardware in
these
> cases. Oversampled sigma/delta? R-2R? What are the
postfilter
> characteristics? I think with these type of stimuli,
differences
> might be audible: even if the same soundcard is used at
different
> rates, the filter should change. The differences in
filter
> characteristics might extend to lower frequencies, where they might
be
> picked up by individuals with good hearing.
>
> Joe.
>
> --
> Joachim Thiemann ::
http://www.tsp.ece.mcgill.ca/~jthiem
Brian C. J. Moore, Ph.D, FMedSci, FRS,
Professor of Auditory Perception,
Department of Experimental Psychology,
University of Cambridge,
Downing Street,
Cambridge CB2 3EB,
UK
Tel. +44 (0) 1223 333574
Fax. +44 (0) 1223 333564
|