Scribd
Upload
34 views

Data Compression: © 2011 SAE Education Ltd. - Subject To Change Without Notice!

Data Compression

Uploaded by

Lucca G. P. Noacco
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
34 views

Data Compression: © 2011 SAE Education Ltd. - Subject To Change Without Notice!

Data Compression

Uploaded by

Lucca G. P. Noacco
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Audio 401

Topic Handout

Data Compression
© 2011 SAE Education Ltd. | Subject to change without notice! INSTITUTE
Table of Contents
Data Compression __________________________________________1
Data Transfer Rate _________________________________________1
Reducing the Data Rate _____________________________________1
Bit Rate Reduction ____________________________________1
Data Compression x Data Reduction _______________________2
Data Compression ('Lossless' Systems) ____________________2
Data Reduction ('Lossy' Systems) ________________________2

© 2011 SAE Education Ltd. | Subject to change without notice!


Data Compression
Digital audio offers:
• Excellent signal to noise ratio (dynamic range of over 90db)
• Bandwidths of 20kHz or greater
• High data rate per channel (e.g. 44.1 kHz / 16 bit = 705 kb/s - Mono)

Data Transfer Rate


Data transfer rate is directly related to the efficiency of storage media and data
networks. Data rate dictates the number of audio channels that may be handled
simultaneously, and how fast files can be transmitted.

Reducing the Data Rate


Use of linear PCM would be best possible alternative.
Two channels of linear PCM require a data rate of 1.4 Mbit/s. This data rate is too
high for most applications. High data rates also equal high cost (storage, transfer
and transmission) so several systems have been developed to reduce data rate.

What is the compression rate between the following streaming rates and and
stereo CD quality audio?
a) 128 kb/s
b) 190 kb/s
c) 320 kb/s

Methods for Reducing the Data Rate


Four possible methods:
• Reducing the sampling rate
• Reducing the bit rate
• Remove redundant data
• Remove irrelevant data

Sampling Rate Reduction


Sampling rate dictates audio bandwidth. Sampling rate reduction is used in
digital broadcast systems and is also found in old samplers and effects
units. In digital broadcast the audio bandwidth may be reduced to 15 kHz
and a sampling rate of 32 kHz is utilized (NICAM / DAB) resulting in a 40%
save in data rate (against 48 kHz). Higher end of frequency spectrum
content is removed. Audio fidelity is affected.

Bit Rate Reduction


Bit rate dictates signal-to-noise ratio Bit rate reduction was used by old
digital equipment Low level signals cannot be encoded properly

© 2011 SAE Education Ltd. | Subject to change without notice! 1


Data Compression x Data Reduction
'Lossless' Systems - A part of the data pertaining the original signal is not
registered (recorded) but may be perfectly represented on replay (data
compression).
'Lossy' Systems - A part of the data pertaining the original signal is not
registered (recorded) becoming irretrievable (data reduction).

Data Compression ('Lossless' Systems)


Redundant data (data that do not carry useful audio information) is not
encoded. Maximum data reduction rates of 2.5:1 dependant on program
material (audio signals are unpredictable / do not spend much time in one
binary set or another / no character set).
The following are examples of lossless techniques:

I - Higher order bits may be removed when the signal to be encoded is


small. Missing higher order bits are reinstated before replay. Maximum data
rate reduction of about 2:1.

II - Reoccurring sample values may be identified, removed and restored at


replay. e.g. Zero crossing points.

III - Prediction Systems Repetitive audio signals are predicted and this
predictions are compared with original signal to be encoded and only the
differences are registered. The decoder will regenerate the same
predicted signal and add changes. Accuracy of system depends on
predictor algorithm. Band splitting improves efficiency over random signals
(multiple predictors).

Drawbacks:
• Encoder / decoder must match and be kept equally updated.
• Possible listener fatigue
• Generation (copy) loss

Data Reduction ('Lossy' Systems)


Irrelevant data (data representing sounds considered to be inaudible in the
presence of other elements of a complex audio signal) is not encoded.
High quality audio may be coded at rates under 100 kbit/s per channel
with minimal effects on quality (192 is considered by many as
indistinguishable from PCM).
Used in digital camcorders, DVDs, laser discs, samplers, sample-based
keyboards, disk- based multi-track recorders, telephone systems, radio and
television broadcast networks, etc.
Coding is based on the non-linear characteristics of human hearing relying
on:

© 2011 SAE Education Ltd. | Subject to change without notice! 2


• Frequency Masking A sound of given frequency and considerable
amplitude will 'mask' sounds in adjacent frequencies. Strongest masking
effects occur when signals occupy the same critical band.
• Temporal Masking (Backward and Forward) A sound occurring within a
short period of time from another sound of considerable amplitude will
be rendered inaudible. Forward masking will be dictated by amplitude
and nature (i.e. envelope) of initial sound and may extend for over 100
milliseconds while backward masking is less strong and extends for
about 5 milliseconds.

© 2011 SAE Education Ltd. | Subject to change without notice! 3

You might also like