A-8
Video/Audio Compression and Decompression Concepts
A.2.1 MPEG Audio Encoding
MPEG audio encoding is intended to efficiently represent a digitized
audio stream by removing redundant information. Because different
applications have different performance goals, MPEG uses different
encoding techniques. These techniques, called Layers provide different
trade-offs between compression and signal quality. The MPEG algorithm
uses the two following processes for removing redundant audio
information:
Coding and quantization
Psychoacoustic modeling
Coding and quantization are techniques that are applied to data that has
been mapped into the frequency domain and filtered into subbands.
Psychoacoustic modeling is a technique that determines the best
allocation of data within the available data channel bandwidth based on
human perception.
The general structure of an MPEG audio encoder is shown in
Figure A.4
.
Figure A.4
Audio Encoding Process (Simplified)
Once audio data has been coded, it may be stored or transmitted
digitally. MPEG provides a framework for use of packet-oriented
transmission of compressed data. In particular, ISO/IEC 11172 defines
formats for digital data streams for both video and audio. The ISO
System Stream format is designed to accommodate both audio packets
and video packets within the same framework for transmission. The data
may be physically delivered in parallel form or serial form. The System
Stream is composed of a sequence of packs, as shown in
Figure A.5
.
Bit Allocation
Processor
(Among Subbands,
Coding, Quantizing)
Frequency
Filter Bank
(Mapping)
Digitized
Audio
Input
Psychoacoustic
Model
Bitstream
Formatter