You are breaking my heart. Are you sure? After mentioning
me twice a propos of nothing in your replies to your other fans, coyly
drawing my attention, you leave me high and dry, just with Pierce and
Audio Guy as the other relay racers. No more literary criticism? No
more hilarious quips?
But wait! You leave behind 18 titles culled a little
carelessly from the annual indices of JAES. You must have been in a
rush because you have one abstract repeated twice.... and one bare
title- no abstract. Still at least we have 16 abstracts. Better than
Pierce who gives titles alone.
So let me try and get "a healthy and sated mind" with a
little help from my friends. You see I don't take offence in a good
cause.
Mercy be! Not a single account of research documenting that
ABX is the proper instrument for us audio consumers to use when
COMPARING COMPONENTS. Remember COMPONENT COMPARISON? Remember us audio
fans? I've been putting it in capitals just to keep minds focused. I
know that some say it isn't good manners. Is that why you ignored it?
What is one to do to impress the topic on you- no underlining, no
bold print or italics in Google postings?
Instead you got: coding systems, testing of MPEG, audio
level monitoring for the blind, noise reduction, transformation of
binomials, verbal and non-verbal elicitation of audio impressions, and
other such burning audiophile issues. There is Mr. Toole, himself
testing:"Subjective measurements of loudspeaker quality" but no
mention of ABX!
You must be a believer in the great dictum: never
underestimate the idiocy of your readers.
There is Prof. Lip****z (he of the "SACD is a
catastrophe")explaining why he believes in "blind or preferably double
blind" testing
but not givimg any results in support.
There is Mr. Clark describing the ABX switch. You put
him in twice.
But you must have been in a rush because you included
as your Nr.8 exhibit a critique by Leventhal of the BIAS in ABX
testing AGAINST recognition of differences. Read it carefully- you
might get it this time. It was hotly debated by ABXers at the time.
Pierce is right. I will not read this stuff. If I
wanted to be thought an expert in psychoacoustics I would have studied
it- not just quoted a hodge-podge counting on no one calling my bluff.
The subject and your references are not of the slightest interest to
me and without any bearing on the subject of suitability of ABX for
untrained, disparate, uninterested audio consumers.
Now if you quoted one, single published audiophile
PANEL ABX test with a positive outcome there would be something to
talk about.
A "test" that has negative results only is not a
"test" for this application: comparing audio components by audio
consumers. Look up any introductory chapter on the methodology of
scientific experiment.
Ludovic Mirabel
The Great Debate: Subjective Evaluation 1170191 bytes (CD aes4)
Author(s): Lip****z, Stanley P.; Vanderkooy, John
Publication: Volume 29 Number 7/8 pp. 482·491; July 1981
Abstract: A polarization of people has occurred regarding subjective
evaluation, separating those who believe that audible differences are
related to measurable differences in controlled tests, from those who
believe that such differences have no direct relationship to
measurements. Tests are necessary to resolve such differences of
opinion, and to further the state of audio and open new areas of
understanding. We argue that highly controlled tests are necessary to
transform subjective evaluation to an objective plane so that
preferences and bias can be eliminated, in the quest for determining
the accuracy of an audio component. In order for subjective tests to
be meaningful to others, the following should be observed. (1) There
must be technical competence to prevent obvious and/or subtle effects
from affecting the test. (2) Linear differences must be thoroughly
excised before conclusions about nonlinear errors can be reached. (3)
The subjective judgment required in the test must be simple, such as
the ability to discriminate between two components, using an absolute
reference wherever possible. (4) The test must be blind or preferably
double-blind. To implement such tests we advocate the use of A/B
switchboxes. The box itself can be tested for audibly intrusive
effects, and several embellishments are described which allow
double-blind procedures to be used in listening tests. We believe
that the burden of proof must lie with those who make new hypotheses
regarding subjective tests. This alone would wipe out most criticisms
of the controlled tests reported in the literature. Speculation is
changed to fact only by careful experimentation. Recent references
are given which support out point of view. The significance of
differences in audio components is discussed, and in conclusion we
detail some of our tests, hypotheses and speculations.
Approximation Formulas for Error Risk and Sample Size in ABX Testing
442116 bytes (CD aes4)
Author(s): Burstein, Herman
Publication: Volume 36 Number 11 pp. 879·883; November 1988
Abstract: When sampling from a dichotomous population with an assumed
proportion p of events having a defined characteristic, the binomial
distribution is the appropriate statistical model for accurately
determining: type 1 error risk (symbol); type 2 error risk (symbol);
sample size n based on specified (symbol) and (symbol) and
assumptions about p; and critical c (minimum number of events to
satisfy a specified [symbol]). Table 3 in [1] pre;sents such data for
a limited number of sample sizes and p values. To extend the scope of
Table 3 to most n and p, we present approximation formulas of
substantial accuracy, based on the normal distribution as an
approximation of the binomial.
High Resolution Subjective Testing Using a Double Blind Comparator
1281885 bytes (CD aes10)
Author(s): Clark, David
Publication: Preprint 1771; Convention 69; May 1981
Abstract: A system for practical implementation of double-blind
audibility tests is described. The controller is a self contained
unit, designed to provide setup and operational convenience while
giving the user maximum sensitivity to detect differences. Standards
for response matching other controls are suggested as well as
statistical methods of evaluating data. Test results to data are
summarized.
Noise Reduction in Audio Employing Auditory Masking Approach 2543054
bytes (CD aes15)
Author(s): Czyzewski, Andrzej; Krolikowski, Rafal
Publication: Preprint 4930; Convention 106; May 1999
Abstract: A new method of noise reduction which exploits some
features of the auditory system is proposed. The noise suppression is
obtained twofold: by rising masking thresholds or by keeping noisy
components beneath these thresholds. The foundations of the method
and some engineered algorithms are described. The way of introduction
of the noise reduction features into an MPEG encoder is demonstrated.
Transformed Binomial Confidence Limits for Listening Tests 468821
bytes (CD aes5)
Author(s): Burstein, Herman
Publication: Volume 37 Number 5 pp. 363·367; May 1989
Abstract: A simple transformation of classical binomial confidence
limits provides exact confidence limits for the results of a
listening test, such as the popular ABX test. These limits are for
the proportion of known correct responses, as distinguished from
guessed correct responses. Similarly, a point estimate is obtained
for the proportion of known correct responses. The transformed
binomial limits differ, often markedly, from those obtained by the
Bayesian method.
Comments on "Type 1 and Type 2 Errors in the Statistical Analysis of
Listening Tests" and Author's Replies 674942 bytes (CD aes4)
Author(s): Shanefield, Daniel; Clark, David; Nousaine, Tom;
Leventhal, Les
Publication: Volume 35 Number 7/8 pp. 567·572; July 1987
Abstract: Not available.
High-Resolution Subjective Testing Using a Double-Blind Comparator
955218 bytes (CD aes4)
Author(s): Clark, David
Publication: Volume 30 Number 5 pp. 330-338; May 1982
Abstract: A system for the practical implementation of double-blind
audibility tests is described. The controller is a self-contained
unit, designed to provide setup and operational convenience while
giving the user maximum sensitivity to detect differences. Standards
for response matching and other controls are suggested as well as
statistical methods of evaluating data. Test results to date are
summarized.
Type 1 and Type 2 Errors in the Statistical Analysis of Listening
Tests 1828932 bytes (CD aes4)
Author(s): Leventhal, Les
Publication: Volume 34 Number 6 pp. 437·453; June 1986
Abstract: When the conventional 0.05 significance level is used to
analyze listening test data, employing a small number of trials or
listeners can produce an unexpectedly high risk of concluding that
audible differences are inaudible (type 2 error). The risk can be
both large absolutely and large relative to the risk of concluding
that inaudible differences are audible (type 2 error). this
constitutes systematic bias against those who believe that
differences are audible between well-designed electronic components
that are spectrally equated and not overdriven. A statistical table
is introduced that enables readers to look up type 1 and type 2 error
risks without calculation. Ways to manipulate the risks are
discussed, a quantitative measure of a listening test's fairness is
introduced, and implications for reviewers of the listening test
literature are discussed.
On the Audibility of Midrange Phase Distortion in Audio Systems
1936662 bytes (CD aes4)
Author(s): Lip****z, Stanley P.; Pocock, Mark; Vanderkooy, John
Publication: Volume 30 Number 9 pp. 580·595; September 1982
Abstract: The current state of our knowledge regarding the audible
consequences of phase nonlinearities in the audio chain is surveyed,
a series of experiments is described which the authors have conducted
using a flexible system of all-pass networks carefully constructed
for this purpose, and some conclusions are drawn regarding the
audible effects of midrange phase distortions. It is known that the
inner ear possesses nonlinearity (akin to an acoustic half-wave
rectifier) in its mechanical-to-electrical transduction, and this
would be expected to modify the signal on the acoustic nerve in a
manner which depends upon the acoustic signal waveform, and so upon
the relative phase relationships of the frequency components of this
signal. Some of these effects have been known for over 30 years, and
are quite audible on even very simple signals. Simple experiments are
outlined to enable the readers to demonstrate these effects for
themselves. Having satisfied ourselves that phase distortions can be
audible, the types of phase distortions contributed by the various
links in the audio chain are surveyed, and it is concluded that only
the loudspeaker contributes significant midrange phase
nonlinearities. Confining the investigation to the audibility of such
phase nonlinearities in the midrange, circuitry is described which
enables such effects to be assessed objectivbely fo their audible
consequences. The experiments conducted so far lead to a number of
conclusions. 1) Even quite small midrange phase nonlinearities can be
audible on suitably chosen signals. 2) Audibility is far greater on
headphones than on loudspeakers. 3) Simple acoustic signals generated
anechoically display clear phase audibility on headphones. 4) On
normal music or speech signals phase distortion appears not to be
generally audible, although it was heard with 99% confidence on some
recorded vocal material. It is clear that more work needs to be done
to ascertain acceptable limits for the phase linearity of audio
components·limits which might become more stringent as improved
recording/reproduction systems become available. It is stressed that
none of these experiments thus far has indicated a present
requirement for phase linearity in loudspeakers for the reproduction
of music and speech.