[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

channel asynchrony

Hi Dave,

When many frequency bands and a variety of delays are used, larger maximum
delays can be tolerated (Greenberg & Arai, 1998; also see Fu & Galvin,
2001).  However, as was pointed out, this does not mean that the delays are
not detectable ("perceptual equivalence").

When speech-modulated tones were used in place of speech bands, and when
the number of channels was limited to two (which allows for the strict
isolation of the effect of asynchrony), it was difficult for us to hear
12.5 ms of cross-channel asynchrony.  Twenty-five ms was clearly detectable
and produced lowered intelligibility scores.  Intelligibility fell toward
zero at 100 ms.  This holds for band pairs drawn from various regions of
the speech spectrum (Healy & Bacon, 2000).

Interestingly, listeners having a hearing impairment have more difficulty
comparing both synchronous and asynchronous across-frequency information,
suggesting that their comparison/integration mechanism may not be as robust
as that of NH listeners (Healy & Bacon, 2002).

Fu, Q-J, & Galvin, J.J., III. (2001).  Recognition of spectrally
asynchronous speech by normal-hearing listeners and Nucleus-22 cochlear
implant users.  JASA, 109, 1166-1172.

Greenberg, S. & Arai, T. (1998). Speech intelligibility is highly tolerant
of cross-channel spectral asynchrony.  Joint Meeting of the ASA and the
International Congress on Acoustics, Seattle, 2677-2678.

Healy, E.W. & Bacon, S.P.  (2000).  Across-frequency timing in speech
perception:  Effective processing of contrasting temporal speech patterns
requires proper synchronous alignment.  JASA, 108, 2603(a).

Healy, E.W. & Bacon, S.P.  (2002).  Across-frequency comparison of temporal
speech information by listeners with normal and impaired hearing.  JSLHR,
45, 1262-1275.

Hope this helps.

Eric W. Healy, Ph.D.
Assistant Professor
Department of Communication Sciences and Disorders
Norman J. Arnold School of Public Health
University of South Carolina
Columbia, SC 29208
(803) 777-1087 voice (803) 777-3081 fax

At 12:00 AM 3/12/2003, you wrote:
Date:    Tue, 11 Mar 2003 12:02:43 +0000
From:    David Dorran <david.dorran@DIT.IE>
Subject: temporal resolution


I've posted this message to the music-dsp list so apologies to anyone who
recieves it twice (should have posted it here in the first place since its
more appropriate).

Consider the case where an audio signal is filtered into subbands and each
subband is passed through a delay (of different duration for each subband)
before recombination of subbands to produce a new signal. What would be
the maximum permissible difference between the maximum and minimum delays
so that the new signal is perceptually equivalent to the original?



Date:    Tue, 11 Mar 2003 11:10:27 -0400
From:    "Dennis P. Phillips, Ph.D." <Dennis.Phillips@DAL.CA>
Subject: Re: temporal resolution

Hi Dave:

Arai & Greenberg (1998) studied speech intelligibility in samples in
which spectral slices had undergone temporal jitter of this general
kind.  My recollection is that asynchronies up to (and sometimes
exceeding) 200 ms could support 50%-correct word recognition.
"Correct word recognition" is, of course, not the same thing as
"perceptual equivalence".  My point is only that the Arai & Greenberg
article might be a good place to start looking for the evidence you
need.  I hope that this helps.  All good wishes,


Arai, T., and Greenberg, S.  (1998)  Speech intelligibility in the
presence of cross-channel spectral asynchrony.  Proc. IEEE Conf.
Acoust. Speech Signal Proc., Seattle, 933-936.
Dennis P. Phillips, Ph.D.
Hearing Research Laboratory
Department of Psychology
Dalhousie University
Halifax, NS, Canada  B3H 4J1
Phone:  (902) 494-2383
Fax:    (902) 494-6585
E-mail: Dennis.Phillips@Dal.Ca
Date:    Tue, 11 Mar 2003 14:00:32 -0300
From:    Regis Rossi Alves Faria <regis@LSI.USP.BR>
Subject: Re: temporal resolution


auditory system is much more sensitive to modulations of any order than
the visual system is;
during my studies with wavelet decomposition/reconstruction of
audio/musical signals, it became very clear that spectral processing
done in some
subbands produces several modulations in the audio signal, which are
sensed as effects in the sound;
I wonder what kind of effects different delays applied to subbands would
cause to the sound, causing the reconstruction to have easily recognized

distortion/modulations even when delays are very short.