reported
by: Dr. Roland K C Tan
Secretary (Term
1995/96)
In the cool and comfortable
KingFisher Room in Orchard Hotel, a total of 6 AES members
with 7 non-members gathered for a very informative lecture on "Audio
Test & Measurement with Audio Precision System Two" on Wednesday, June
12, 1996 at 9.30pm. This is the section's third seminar this year
and it was held at the heart of the busy city in Singapore. Mr. Steve
Peterson, International Sales Manager of Audio Precision Inc.,
Oregon, USA and the presenter of this seminar, shared his wealth of superb
knowledge in the field of audio analysis with the audience comprising mainly
of engineers. This lasted for almost 4 hours, stretching the entire morning
lecture till way past lunch hour in the afternoon.
Peterson, with over 25
years of experience in test and measurement, first gave a brief introduction
of the new AP's System Two audio analyser. Launched only recently, the
features in System Two aimed to meet higher performance requirements and new
needs of audio engineers.
By projecting the AP software
environment on screen, Peterson was able to show the audience many of
the useful and user-friendly features found in the new APWIN software that
come with System Two. One such major feature is the ability to perform
multi-tasking. This allows multiple graph, bargraph, generator, and
analyzer, among others, to run simultaneously and display on screen. With
APWIN for Windows 95, a true 32-bit operating system replaced the old DOS
software. This effectively increased the processing power and permits a
faster FFT screen updates to follow program material and other rapidly
changing signals besides other enhancement.
After having introduced the
equipment, Peterson then proceeded to talk on the technical topics
such as the fundamental of digital audio serial interface. Interface
standards such as the AES/EBU format as well as the SPDIF formats were
discussed. He later explained how interface tests can be conducted by
injecting jitter signal. The objective of performing such test is to
determine whether signal from a digital audio device meets standards and is
thus acceptable to another device. Sinusoidal or random jitter signal can be
generated using System Two. The rise and fall times, and amplitudes of the
pulse train output can also be varied. The effect of jitter distortion can
then be analysed using FFT of jitter or eye patterns plot.
Before a 15 minutes coffee break,
the testing of high resolution A/D and D/A converters was described. Test of
up to 20-bit resolution A/D and D/A converters with good accuracy is
possible using multi-tone test signals. Traditional THD measurement method
may not be feasible especially for signals at higher fundamental frequency
tone of say, 15 kHz. The second harmonic at 30 kHz would be beyond the
bandwidth of normal human hearing! To ensure accurate results, Peterson
also stressed the importance of adding dither signal before performing any
linearity tests. However, he said, that it is equally important to remember
disabling the dither signal source before performing any THD measurements.
During the second half of the
session, Peterson continued with his lecture and discussed about
synchronous FFT and multi-tone testing techniques for broadcast systems and
tape. By using the technique of synchronous FFT the usual problems of
windowing and poor frequency selectivity, associated with standard FFT
short-time spectral analysis technique, can be eliminated. One member
brought up a suggestion of exploring other higher resolution spectral
estimation techniques for audio analysis purposes. Techniques such as the
parametric methods (ARMA, MA, Prony's) or the maximum likelihood (ML) method
were mentioned. These methods are commonly found in other area of
applications especially in speech processing and communication analysis. On
audio testing using multi-tone signals, Peterson demonstrated to the
audience with the equipment he had set-up earlier on how THD,
intermodulation, interchannel phase, interchannel balance, and crosstalk
interference in stereo systems can be measured.
On the testing of low bit-rate
perceptual codecs, conventional test methods are simply far from adequate.
To realistically test such codecs, the test signal must have the equivalent
complexity as compared to an actual music or voice source. A more complex,
music-like test signal is required to stress the codecs in this case.
Instead of using a single sinewave test signal, synchronous multi-tone test
signal can be generated as complex as desired. With System Two, each
sinewave tone signal can be selected at different frequencies and
amplitudes. In addition, the system allows one to generate and overlay
composite masking curve and also analyse distortions across critical bands.
The objective results obtained this way would be closer to that of the
actual human hearing system.
During the question and answer
session, a question was raised concerning the possibility of integrating
speech analysis features in future AP's System audio analyser since
multimedia sound system includes both audio and speech sound signals. In
reply to another question regarding "waterfall" plot feature, Peterson
revealed that this is already included in later version of the APWIN
software.
The Singapore Section would like
to thanks the Managing Director of TME Systems Pte Ltd, Singapore, Mr.
Ronald Soo, for his generous support and sponsorship. Also to Mr.
Thomas Chen, Sales & Application Engineer and also the main organiser,
for extending his invitation to all AES members. |
