reported
by: Mr. Jibby Jacob
Secretary (Term
1996/97)
On Thursday, 24 April 1997
at 7.45 pm at the Ngee Ann Polytechnic - Department of
Electrical Engineering's staff conference room, AES Singapore Section's
Chairman, Dr. Roland K. C. Tan conducted a 1-hour tutorial
session to 4 members and 9 guests on "Bit-Rate Reduction
Techniques in Digital Audio".
| Traditionally, waveform
coding technique have been applied successfully both in the field of
speech and audio signal processing. The main objective of Dr. Tan's
talk was on incorporating parametric coding technique as an alternative
to waveform coding techniques for applications such as internet
communications that use low bit-rate channels. The performance of APCM,
ADPCM, and ADM is first compared against LPC parametric doing technique,
to determine their compatibilities with sub-band coding.
The proposed MASC algorithm
uses parametric coding combined with multi-pulse LPC and sub-band
coding. |
Adaptive perceptual error
F-weighting filter is incorporated in the LPC optimization loop to motivate
the excitation pulse locations. A further efficiency enhancing strategy
allocates the number of excitation pulses within the sub-band set based upon
the instantaneous power. It was demonstrated that these techniques could
achieve improved subjective sound quality within a constrained bit budget.
Dr. Tan then gave a sound
demonstration to the audience using pre-recorded musical materials that have
been processed with the proposed MASC algorithm at a reduction ratio of more
than 6:1. This is equivalent to coding at less than 3-bit/sample in the
linear PCM format. Both classical as well as pop music materials were used.
MASC codec could offer high sound quality without discarding samples based
on noise masking theory found in some existing algorithms. During the
question-and-answer session, several questions were raised. In response to
one of the questions, Dr. Tan agreed that there should be no
reduction in bit-rate associated with current 16-bit or even higher density
(24-bits/ 96 kHz) linear PCM format as far as archive materials are
concerned. This is especially so since the availability of higher density
DVD storage technology.
However, to be able to
substantially reduce the high data rate associated with the newer format has
strong economic and design consequences in other applications such as the
storage and transmission areas. For examples, the higher resolution quality
can be "preserved" onto current CD storage medium at 16-bits/44.1 kHz
without compromising the total playing time of around 74 minutes.
Alternatively, musical signals can be encoded and transmitted through the
ISDN channels at a compressed rate of 128 kb/s. In Digital Audio
Broadcasting (DAB), more high-quality stereo or even multi-channel
programmes can be multiplexed into the finite electro-magnetic spectrum.
|
