RoadRunner
U S E R S G U I D E
34
Algorithm
Sample
rate
Bit rate
Bandwidth
Delay
Cascade
Layer III
32 kHz
56/64 kb/s
15 kHz
long
poor
4
Layer III
48 kHz
56/64 kb/s
15 kHz
long
poor
MUSICAM
48 kHz
112/128 kb/s
20 kHz
moderate
excellent
4
Layer III
48 kHz
112/128 kb/s
20 kHz
long
fair
G.722
is an early compression algorithm based on ADPCM (Adaptive-
Predictive Pulse Code Modulation). The delay of this algorithm is very
low, and is ideal when talk-back or two-way, on-air conversations are
required. Although G.722 has limited bandwidth and audible artifacts,
it is quite acceptable for most voice applications. G.722 does not
cascade (re-compress) well and artifacts build up quickly when a
decoded bit stream is re-encoded.
MUSCIAM
, MUSICAM USA’s enhanced version of ISO/MPEG Layer
II, offers the best possible combination of fidelity, moderate delay and
cascading ability at all bit rates. For best possible sound when using
lower bit rates (56 and 64 kb/s), use 24 kHz sampling if it is supported
by the far-end codec. At 56 or 64 kb/s using 24 kHz sampling,
MUSICAM delivers a bandwidth of 10.5 kHz with exceptional signal-
to-noise figures and minimal artifacts. At this bit rate MUSICAM is
ideal for all voice applications and delivers satisfactory performance for
many music applications.
MUSICAM cascades well and the sound quality does not degrade with
post-processing and equalization. At 112 or 128 kb/s, MUSICAM is
transparent and provides a bandwidth of 20 kHz with exceptional
signal-to-noise performance.
Please remember that MUSICAM is a fully compatible enhancement to
the ISO/MPEG Layer II standard algorithm. All enhancements are to
the encoder; the decoder is defined by the international Standards
organizations, and has been left unchanged. Therefore, when receiving
audio from a standard Layer II codec, the audio you hear will be Layer
II, but the audio heard at the far-end will be MUSICAM. MUSICAM
encoding makes
any
Layer II decoder sound better.
!
note:
MUSICAM
enhanced MPEG
Layer 2 at 128
kb/s, 48 kHz
sampling is the
recommended
algorithm for all
professional voice-
over applications.