From: malcolm@interval.com (Malcolm Slaney)
Date: Mon, 22 May 1995 21:23:00 -0700
Subject: Spatial Hearing and Sound
Message-Id: <abe704fb00021004a125@[199.170.108.19]>

We've got two application talks coming up at the CCRMA Hearing Seminar.
How is it that we can use the psychoachoustics we know to build useful
devices?

Matti Karjalainen has been visiting CCRMA for the last year.  His talk this
week will focus on spatial hearing and sound quality measurements.  A week
from Thursday Bill Putnum will be reviewing the literature and experiments
in sound quality measurements.

Possibly the biggest use of psychoacoustics is in audio coding.  Generally
some sort of auditory model is used to measure the important pieces of the
sound.  A related technique is the use of an auditory model to evaluate
sound quality.  Many of the devices we use modify the sound waveforms.  How
do we decide how much these modifications have changed the perception of
the sound?   This is a theme for the next two talks at the CCRMA Hearing
Seminar.

This coming Thursday Matti will be talking about estimating spatial
direction from a binaural recording.  What are the important clues and what
does it tell us about how humans perceive direction.  He'll also describe
his work on sound quality measurement.

        Who:    Matti Karjalainen (Helsinki University of Technology)
        What:   Auditory Models for Spatial Hearing and Sound Quality
        When:   Thursday May 25th at 11AM
        Where:  CCRMA Library (Top Floor of the Knoll) at Stanford

See you at CCRMA!

-- Malcolm

--------------

Applications of Auditory Modeling to Spatial Hearing and Sound Quality
Measurements.

Matti Karjalainen

The talk will cover two application-oriented studies on auditory modeling,
both from the Helsinki University of Technology, Acoustics Lab. The first
is an experiment on the ability of an artificial directional
listener--consisting of a dummy head, a simple binaural auditory model, and
a neural network--to learn to estimate the direction of a sound source. It
was found that in simple environments the system outperforms humans but the
ability to generalize for complex sound environments is poor. This work was
originally published in the IEEE ICASSP-93 proceedings.

The second topic describes early work (1981-86) on psychoacoustically based
auditory modeling in order to measure sound quality degradation due to
various types of distortion. Similar approaches have recently become
popular due to the need to evaluate the quality of modern audio coding
techniques. The model in this study was applied to compute a spectral
distance measure between the original undistorted signal and the output of
a distorting channel. The technique was applied primarily to speech signals
and it allows for using any signal as a test signal for distortion
measurement.