Subband Coder
Speech-Compression
Algorithm
by Atlanta Signal Processors, Inc.

Software Overview

The ASPI Subband Coder is a high-quality medium-bit-rate coder. It employes Adaptive Pulse Code 
Modulation (APCM) for low quantization error over a wide dynamic range. It utilizes five octave 
subbands to compress and reconstruct speech with an extremely high level of intelligibility. It is 
configured for 16-kbps data rate and optimized for speech sampled at 8,000 samples/second.
Atlanta Signal Processors' Subband Coder can be implemented full-duplex with a single 20-MHz 
TMS320C10, utilizing only 80% of available CPU cycles. Or up to four full-duplex (or eight 
half-duplex) subband coders can be run concurrently on one TMS320C30. Other configurations have 
been developed for use at rates from 9.6 to 19.2 kbps. The modular design allows exceptional 
flexibility to meet specific requirements.
A variable-rate Subband Coder for the first- and second-generation TMS320 processors is also 
available. The Variable-Rate Subband Coder dynamically adjusts the number of bits assigned to each 
of its bands according to the energy of the signal in that band. It does this by selecting among a 
range of available quantization schemes for each band. The quantizers in each band are adjusted 
independently of each other. The resulting system offers the quality of a full 16-kbps subband 
coder, but at a greatly-reduced average data rate.
This coding system works well in situations where a fixed data-rate channel is not being used, such 
as for voice store-and-forward, voice mailbox, or statistical channel allocation.
A version of the Subband Coder is also available that permits up to four full-duplex (or eight 
half-duplex) subband coders to be run concurrently on one TMS320C3x.

Features and Benefits

•	Low-computational requirements
•	Far less sensitive to transmission bit errors than other vocoders
•	Can handle multiple simultaneous voices with virtually no increase in de-gradation

Processor and System Specifications

•	Texas Instruments TMS320C1x, TMS320C2x, TMS320C3x, or TMS320C5x TMS320C3x, 16-kbps, sample-rate 8 
kHz:
Program memory: 1,668 words
Data memory: 960 words
CPU loading (dynamic memory): 
38.5% encoding

Usage Limitations or Performance Considerations

•	The Subband Coder is currently optimized for speech in the 300 Hz to 3 kHz band.

Algorithm Verification

•	Speech quality measurements results:
Diagnostic Rhyme Test Score:
94 @ 16-kbits data rate
•	Modified Rhyme Test Score: 
83 @ 9600-bit data rate
•	The Subband Coder has been extensively used by customers world-wide.

Availability

•	ASPI's Subband Coder algorithm is available under a nonexclusive, worldwide, non-transferable 
license for a one-time fee. This license allows the algorithm to be incorporated into any of your 
company's products. Fully commented source code for the Subband Coder algorithm is delivered along 
with demonstration software. Documentation includes subband coder theory.
•	A demonstration tape is available that contains samples of speech encoded and reconstructed by 
the ASPI Subband Coder. These segments demonstrate the Subband Coder's performance on male and 
female voices under controlled conditions with low-background noise and scripted speech.

Company Background and Contact Information

Atlanta Signal Processors was founded in 1981 by Ronald W. Schafer, Thomas P. Barnwell III, and 
Russell M. Mersereau, three Professors of Electrical Engineering at the Georgia Institute of 
Technology with over 80 years of experience in the field. The company has focused its efforts on 
developing tools for professional signal-processing design engineers and now offers digital 
signal-processing hardware and software products to DSP designers worldwide.
Contact:	Atlanta Signal Processors, Inc.
		1375 Peachtree St. NE, Suite 690
		Atlanta, Georgia  30309-3115
		(404) 892-7265
		Fax: (404) 892-2512
		e-mail: aspi@aspi.com