1464 Chapter 39
sample of audio is represented by a number indicating
its place in a range of equal-sized amplitude steps. If
eight bits were being used to represent the audio level,
there would be 2^8 or 256 equal-sized amplitude steps
between the smallest level that could be represented and
the largest. The smallest amplitude change that could be
represented is exactly the same at the low-level portion
of the range as at the highest-level portion. This is
important to understand since not all digital audio uses
such a linear representation. For example, often tele-
phone calls are encoded using nonlinear techniques to
maximize the speech quality transmitted using a limited
number of bits. In professional audio we generally use
larger numbers of bits, usually in the range of 16 to 24,
that allow excellent performance with linear representa-
tion. Linear representation makes it easier to build high-
quality converters and signal processing algorithms.
The bits that make up an audio sample word are
represented in two’s complement form, and range from
the least significant bit (LSB) that represents the
smallest possible amplitude change, to the most signifi-
cant bit (MSB) that represents the polarity of the signal.
AES3 adds a considerable amount of structure
around the basic sequence of bits described above in
order to allow clock recovery from the received signal,
provide a robust signal that is easily transmitted through
paths of limited bandwidth, and provide for additional
signaling and data transmission down the same path.
Each of the two audio channels that can be carried
by an AES3 signal is formatted into a sequence of two
subframes, numbered 1 and 2, each of which follows
the following format.
The following information is based on AES3-2003.
It is always advisable to obtain the latest revision of the
standard.
39.2.1 Subframe Format
First, additional bits are added before and after the
digital audio to make a subframe of exactly 32 bits in
length. The bits are transmitted in the sequence shown
from left to right, Fig. 39-3.
If the audio data contains 20 or fewer bits, subframe
format B is used. If the audio data contains 21 to 24 bits,
subframe format A is used. In either case if data
containing less than 20 or 24 bits is used, extra zeros are
added to the LSB to bring the total number of bits to 20
or 24. Since the data is in two’s complement form, it is
important that the MSB, representing the signal polarity,
always be located in bit 27.
The preamble is used to indicate if the audio to
follow is channel one or two, and to indicate the start of
a block of 192 frames.
If 20 or less audio bits are carried, then AES3 allows
4 bits of other data to be carried by the AUX bits.
The validity bit is zero if it is permissible to convert
the audio bits into analog, and one if the conversion
should not be done. Neither state should be considered a
default state.
The user bit may be used in any way for any
purpose. A few possible formats for using this bit are
specified by the standard. Use of one of these formats is
indicated by the data in byte one, bits 4 through 7 of the
channel status information. If the user bit is not used it
defaults to zero.
The channel status bit carries information about the
audio signal in the same subframe, in accordance with a
scheme that will be described later.
The parity bit is added to the end of each subframe
and is selected so the subframe contains an even
number of ones and an even number of zeros. This is
called even parity. It allows a simple form of error
checking on the received signal.39.2.2 Frame FormatA subframe from channel two follows a subframe from
channel one. The pair of subframes in sequence is called
a frame.
In normal use frames are transmitted at exactly the
sampling rate.Figure 39-3. AES3 subframe format. Note that the first bit is called bit 0.0 3 4 27 28 31V Validity bit
U User data bit
C Channel status bit
P Parity bitAUX Auxiliary sample bits
0 3 4 27 28 31Preamble LSB 24-bit audio sample word MSB V U C PPreamble AUX LSB 20-bit audio sample word MSB V U C P