Thread Subject: Re: Amplification and Research

Note

This archival content is maintained by WebAIM and NCDAE on behalf of TEITAC and the U.S. Access Board . Additional details on the updates to section 508 and section 255 can be found at the Access Board web site.

From: Michaelis, Paul R. (Paul)
Date: Sun, Mar 25 2007 8:55 AM

I believe that the issue of intelligibility for hard-of-hearing users is
extremely important. I'm struggling with how to write requirements for
Section 508 that would achieve what we want.

A problem is that many of the techniques that are commonly used to
assess telecom voice quality are oriented toward differentiating among
products that, for most people, are fully intelligible. The goal is to
measure perceived quality, rather than intelligibility. This is why
these tests are described as "subjective" with the results reported as a
"Mean Opinion Score." Although the conduct of such tests is a standard
part of our product development process, I have absolutely no idea
whether our measures of perceived quality would correlate highly with
intelligibility among listeners who are hard-of-hearing.

Tests for intelligibility do exist. Examples of procedures that rely on
human assessment include the Diagnostic Rhyme Test and the Modified
Rhyme Test. Unfortunately, given the wide variation among people who
are hard-of-hearing, it would be impractical for me to hire a
statistically valid sample of hard-of-hearing listeners in order to do
DRT and MRT tests for each of my products. The tests that measure the
physical properties of the speech -- for example, the amplitude and
signal-to-noise ratio of the speech at different parts of the acoustic
spectrum -- work pretty well as a predictor of intelligibility for
people with normal hearing. Are these tests valid for people who are
hard-of-hearing?

A bigger problem is that the voice quality in a VoIP system can vary
tremendously depending on factors that are outside the vendor's control,
and beyond the ability of a typical contract officer to assess. For
example, an IP network that has excellent voice quality under normal
conditions may have unacceptable quality when the system is heavily
loaded with voice traffic -- as might happen during an emergency or
during times of the day when large numbers of conference calls tend to
be scheduled. This is because higher levels of traffic can cause
increased packet loss, packet jitter, and packet latency, along with a
need to shift from uncompressed G.711 audio encoding to a compression
technique such as G.729. I have no idea how to write an acceptable
requirement that addresses this issue.

Despite the complexity of these problems, I'm sure we can do a better
job for hard-of-hearing users than simply to require support for
increased amplification. I'm looking forward to hearing people's
suggestions.

-- Paul Michaelis








-----Original Message-----
From: Jim Tobias [mailto: = EMAIL ADDRESS REMOVED = ]
Sent: Sunday, March 25, 2007 7:15 AM
To: 'TEITAC Telecommunications Subcommittee'
Subject: Re: [teitac-telecom] Amplification and Research

David, you are quite right to differentiate between amplification and
clarity, of course. A louder bad signal is not a better signal. Except
in rock music, of course.

Clarity itself is made up of several different characteristics, as Paul
has already responded. I readily admit to not being an expert in this
area.
But I do know that there is a set of industry measures (or at least
testing protocols and metrics) for intelligibility that are commonly
applied to telephones. In fact, there's a whole family of them, P.800
from ITU, and another one from ITU, G.107. Most (all?) of these use
subjective testing: a bunch of people listen to the audio and rate it on
a scale of intelligibility. (Who is in that bunch, of course, is an
interesting question -- you have to wonder if they automatically rule
out anyone with any hearing loss!) The answer is a number. For the
most common metric, MOS (great to have an acronym, but all it stands for
is "mean opinion score"), a rating above 4.0 out of 5.0 is considered
good.

TEITAC or the Access Board could decide to research this issue to find
out:

1. if any other worldwide accessibility regulations use one or more of
these measures 2. if any US regs for any purpose use one or more of
these measures 3. if procurement ever includes any such requirements 4.
which, if any, of the metrics are commonly used by vendors

and perhaps other points about the utility and feasibility of adopting a
metric. One interesting question would be to map the metric onto the
world of hearing loss. For example, how many people with what kind and
degree of hearing loss can detect a difference between a 3.5 and a 4.5
MOS.


***
Jim Tobias
Inclusive Technologies
+1.732.441.0831 v/tty
+1.908.907.2387 mobile
skype jimtobias

-----Original Message-----
From: Baquis David [mailto: = EMAIL ADDRESS REMOVED = ]
Sent: Monday, March 12, 2007 4:03 PM
To: = EMAIL ADDRESS REMOVED =
Subject: [teitac-telecom] Amplification and Research


I have skimmed an archived thread of discussion about amplification as
well
as listened to part of today's call. Here are my thoughts as you
prepare to
present your work-to-date to the TEITAC:

Remember that you can recommend that the Access Board research an issue.
We would likely do that by funding a contractor. Your recommendations
need
not be limited to comments about the Standards/Guidelines.

Would you like research on what is sufficient for hearing access over
the
telephone? I know, of course, that the answer to this is complex since
there
are different types of hearing loss, telephones, situations, etc. And
yet,
you are recommending design standards to help with such access.

For a second, let's set aside the legal issues as well as the burden and
begin with an understanding of what people who are hard of hearing
really
need for accessibility, if we can. In this paragraph, I am not
addressing
what problems we might encounter if we try to accomplish this. I am
simply
asking what is truly helpful.

* When is 20 dB, 25dB, up to 40dB beneficial? Are these demands
based solely on anectodal reports? If there is a science to this,
shouldn't
we discover it? If it is already known, could it be referenced as a
basis
for rulemaking? If we do not know this with research-based certainty,
wouldn't you like to recommend that it be investigated?
* How does clarity interrelate with amplification? If clarity is
improved, does that partially reduce the need for high gain? Does the
subcom need technical information on what specific elements make up
clarity?
* Is a separate standard for clarity needed?
* Any other questions regarding maximizing residual hearing to add
to this list?

Looking back to Brenda's email from November 6, 2006, I see that HLAA
commissioned lab research to address distortion. Would it be helpful to
the
TEITAC to know of other research, such as this, which could help inform
rulemaking? Brenda, do you want to post that research on the wiki (or
ask
Tim to do it). If the TEITAC feels that distortion caused by
amplification
is a non-issue then perhaps that should be stated clearly as a point of
agreement.

Since we are looking at both 255 and 508, consider the impact not only
on
the government customer (508) but on the individual consumer (255).
Of course there is the impact on manufacturers too, so you may want to
lay
this out in an easy to follow chart - with both positive and negative
impacts included.

With regard to Paul's comment on battery drain to provide amplification
at
higher levels, you may want to consider commenting on wireless and
wireline
separately if that makes a difference in respective impact. I am
imagining
not only economic impact, but consumer satisfaction with a phone that
may
not work as well.

However, I am also imagining that the past does not equal the future.
What feels impossible now may become possible in the future with
improvements in batteries, processing, etc.

Oh and don't forget VoIP and network-based amplification when you
consider
possibilities and impacts. Do you imagine old technology or new
technology
when discussing this? Must the power for the volume always be contained
inside the phone set? How far away are we from the day when our
preferred
hearing interface can be profiled -- and we can virtually bring that
profile
with us from phone to phone (or multiple profiles depending on changes
in
environment, congested hearing, etc)? And what would be the hardware
requirements to support a network-based solution?

David

David Baquis
Accessibility Specialist
U.S. Access Board
1331 F Street, NW, #1000
Washington, DC 20004
800-USA-ABLE; (202) 272-0013 (voice)
www.access-board.gov; = EMAIL ADDRESS REMOVED = "Leading the way to
excellence in accessibility"

WebAIM is an initiative of:
Center for Persons with Disabilities (CPD) Utah State University