We use Optical Character Recognition (OCR) during our scanning and processing workflow to make the content of each page searchable. You can view the automatically generated text below as well as copy and paste individual pieces of text to quote in your own work.
Text recognition is never 100% accurate. Many parts of the scanned page may not be reflected in the OCR text output, including: images, page layout, certain fonts or handwriting.
on a wide variety of program material, with a series of frequency range restriction filters giving the frequency characteristics shown in Fig. 1. These were chosen as being representative of a system of the highest caliber (3 db. down at 30 and 9,000 cycles) ; an average quality radio receiver (3 db. down at 160 and 3,500 cycles), and a receiver midway between these limits (3 db. down at 70 and 6,500 cycles).
Representative Audience Chosen
Particular care was taken to eliminate the possible effects of such secondary factors as order of presentation, appearance of the loudspeaker, effect of signal lights, etc. High-quality disc recordings have the advantage of consistency and repeatability, and as a separate test indicated that the results were not appreciably modified by their use, discs were employed for the majority of the tests.
The audience chosen (about 500 people) was representative of a wide variety of possible tastes. It might be expected that a specialized group of listeners, such as those owning FM receivers, would display a more elevated taste than the ordinary listener, who might reasonably be expected to be accustomed to something worse than the narrow-range reproducer.
High-quality reproduction is one of the main advantages claimed for FM technique and steps are taken to make certain that the receivers meet this claim. The even more specialized group of listeners composed of professional musicians accustomed to hearing music, etc., "in the natural," might be expected to exhibit an even more refined taste. Separate tests were therefore taken on a group of people owning FM receivers and on a group of musicians.
A wide variety of program material was made available, separate tests being taken of each type.
The results were surprising, to say the least. When wide range is compared directly with the medium range, the latter scores in eight out of the eleven comparisons, wide range failing to make
+ 5
o
-5
-10 -15 -20 -25 -30 -35
SO 100 S00 IK 2 5 10000
Frequency Cycles per second
Wide — — —-Medium Narrow
/
y
r^
">.
/
•
\ \
1
\ 1
1
/
1
1
/
•
1
'
"or ,
S o
o
i§
(V)
PC
-ro
-75 -20
-25 '30
o
o
XM
Q
o o
o
o
o o
f-requencyj cycles per sec.
o o
Q
o
o
FIGURE 2. Overall frequency characteristic (adjusted for best aural results).
FIGURE 1. Frequency characteristics used for quality judgment.
a decisive score at any point. On the tests where wide range might be expected to score (musicians and FM listeners) , it is in fact most decisively rejected, the musicians being particularly single-minded in their rejection. On comparing wide range with narrow range, the result is found to be almost as emphatic, narrow range scoring in nine out of the eleven comparisons!
Quality Adherents 'Miffed'
The results of the comparison between medium range and narrow range does a little to restore the engineer's selfconfidence, in that medium range scores in seven out of the eleven tests, with one match drawn, and a large percentage of the audience having no preference. The results cannot fail to surprise the quality enthusiast, particularly as it should be noted that the usual accompaniment to "high fidelity" — surface noise, monkey chatter, whistles, etc. — were absent.
In another field there is further evidence of the advantages of operating a sound reproducer with a drooping frequency characteristic at the upper end. Fig. 2 indicates the overall frequency characteristic recommended by the Academy of Motion Picture Arts and Science after careful comparison tests in several theatres. Our own experience confirms this choice.
In the space available for an article of this nature it is not possible to quote further evidence in favor, so we will conclude this section with the assumption that after every effort has been made to reduce distortion, noise, etc., to a minimum, the human ear still prefers the frequency range to be restricted at the upper end of the scale.
A critical review of the evidence quoted in favor of this viewpoint will immediately raise some points on which the evidence is not so conclusive as is really desirable. In the test results quoted there were no distortion or frequency characteristics taken on the
specific set of speakers used for the test, though they were stated to be units of the highest quality, and a group of expert listeners agreed on the excellence of their performance.
After making a close study of the human hearing mechanism, the writer feels surprised that 5 per cent distortion is detectable, as an addition to the really serious distortion occurring within the ear. In this respect it is worth noting that the tone quality (harmonic content) of almost every instrument varies enormously with loudness in normal playing, yet little impression of unpleasant distortion is given.
If our electro-acoustic equipment is really distortionless (i.e., if 1 or 2 per cent is of little consequence) one is immediately led to consider whether the same preference should be shown for restricted top response if electro-acoustic equipment were not involved as a link between instrument and listener.
Instrument Tone Characteristics
Now it would be ridiculous to assume that the frequency response of a violin, for example, was the only factor that determined tone quality, but there is no doubt that it is a very important factor. This question of violin tone quality has been the subject of a very thorough (but as yet uncompleted) investigation by a group at Harvard University.
The results thus far are illuminating. Good violins are characterized by a response which falls off fairly rapidly above 4,000 cycles per second; while the bad models have a response which is well maintained up to, and beyond, 6,300 cycles. At the low-frequency end the good models are again characterized by a well maintained output; while the response of the bad models fall away in much the same manner as a poor amplifier.
In a good instrument the lower tone
will thus be rich and full with prominent
partials (or harmonics) ; while the upper
notes will tend towards pure tones, or
(Continued on page 27)
INTERNATIONAL PROJECTIONIST
February 1947