Acoustic Echo Canceller
for the TMS320C5x
by Signals and Software Limited

Software Overview

DSP software which implements an acoustic echo canceller. This allows loudspeaking 
video-conferencing terminals to use full-duplex audio by cancelling the feedback that occurs 
between the loudspeaker and microphone(s). The echo canceller has been designed for use with 
G.722-specified wideband (0–7 kHz) audio sampled at 16 kHz. A narrowband 8-kHz sampled interface is 
also selectable.

Features and Benefits

The software has been extensively tested and optimized in video-conferencing environments to give 
excellent subjective performance.
Algorithm:			24 sub-band structure with oversampling, complex adaptive FIR filters. Additional echo 
control by adaptive echo attenuation.
Echo tailspan:		200-ms to 100-ms lengths cancelled depending on frequency band.
Bandwidth:			0 to 7 kHz (wideband mode) or 0 to 3.4 kHz (narrowband mode).
Delay:			60 ms in microphone path.
Cancellation depth:	30–35 dB, adaptive filters only, up to 75 dB total, echo attenuator in place.
Convergence time:		1–2 s to -30 dB (noise)/3–5 s to -20 dB (speech), adaptive filters only, <10 ms 
to -75 dB, adaptive filters plus echo attenuator.

Processor and System Requirements

Acoustic
Echo Canceller
Program Memory (Words)
Data Memory
(Words)
Processing Load (MIPS)
Wideband
10 k
7.6 k
28
Narrowband
8 k
5 k
20
The software has been primarily designed to run on a 35-ns (28.5 MIPS) TMS320C53, with some 
external memory, but is portable to other 'C5x processors. The processing load is likely to vary 
slightly with different memory configurations. The program size could also be significantly reduced 
with only a small increase in processing load.

Usage Limitations or Performance Considerations

The software is in the form of two subroutines, one for initialization and one for processing. The 
initialization routine is called normally only on DSP reset. The processing subroutine operates on 
21 sample frames of data. Three sample buffers are provided, for loudspeaker signal input, 
microphone signal input, and loudspeaker signal output. A single control word can be programmed for 
echo canceller on/off and wideband/narrowband features.
Applications include echo control for roll-about, desktop videoconferencing units, and ISDN 
loudspeaking telephones.

Availability

Now, under licence, for a one-off payment and/or royalties depending on the commercial application. 
An Application Note is available. Support consultancy for code integration is also available.

Company Background and Contact Information

Signals and Software Limited (SASL), based in Harrow, Middlesex, UK is a design consultancy 
specializing in the area of Digital Signal Processing (DSP). From concept and algorithm design 
through to real-time DSP implementation, SASL is able to offer its clients fast and cost-effective 
solutions to their DSP needs. Services include: feasibility studies, DSP software to order, DSP 
research and algorithm design, computer simulations (C or PASCAL), and hardware design. In support 
of these services, SASL offers a range of "off-the-shelf" software that includes audio/speech 
coding, modems, acoustic echo cancellation, video coding, and general telecommunications functions. 
Key technology areas include: GSM/PCN mobile comms, video conferencing, video telephony, and 
PSTN/ISDN voice and data transmission.

Contact:	David Morley
		3 Jardine House
		Bessborough Road
		Harrovian Business Village
		Harrow, Middlesex, HA1 3EX
		United Kingdom
		+44 (0) 181 426 9533
		Fax: +44 (0) 181 869 1182
		e-mail: davem@sasl.demon.co.uk