FED-STD 1016
February 14, 1991
FEDERAL STANDARD
TELECOMMUNICATIONS: ANALOG TO DIGITAL CONVERSION OF RADIO
VOICE BY 4,800 BIT/SECOND CODE EXCITED LINEAR PREDICTION (CELP)
This standard is issued by the General Services Administration pursuant to
the Federal Property and Administrative Services Act of 1949, as amended.
1. SCOPE
1.1 Description. This standard specifies
-related requirements for the conversion of
analog voice to a 4,800 bit/s digitized form for digital radio transmission by a method known as Code
Excited Linear Prediction
and the reverse process, the synthesis of analog voice from
4,800 bit/s
voice. In addition to digital radio applications, CELP is also very suitable
for encrypted telephone use and other applications wherein voice must be digitized prior to
encryption.
1.2
This standard is to facilitate interoperability between radio telecommunication facilities
and systems of the Federal Government.
1.3
This standard shall be used by all Federal departments and agencies in the design
and procurement of all radio equipment employing 4,800 bit/s digitized voice.
2. CONVENTIONS AND DEFINITIONS
a. Frame. A CELP frame is 144 bits in length. A frame interval is 30 ms
percent in
duration and contains 240 voice samples (8,000 samples/s).
b.
A CELP
is 1/4 the length of a frame.
Thus, a
is
7.5 ms
percent in duration and contains 60 voice samples.
REQUIREMENTS
3.
3.1 Voice
3.1.1 General Description. S i n c e Code Excited Linear Prediction
is
by-synthesis type of technique, voice synthesis is described first. As shown in the diagram below,
CELP synthesis involves the excitation of a parameter-adjusted Linear Prediction Filter
the sum
of gain-scaled codes selected from a fixed,
derived "stochastic code book"
adaptive "code book," utilizing parameters transmitted in a 144-bit frame structure.
Codes
Adaptive Codes
...
●
●
Linear
●
➤
Prediction
●
●
●
,
LSPS
O