Talking Bank Machines

by Len Fowler

From the Editor: Len Fowler is co-founder and Chief Operating Officer of T-Base Communications of Ottawa, Canada.

Let me begin by stating that I believe everyone in a democratic society is entitled to be provided with opportunities to participate and contribute. I believe that acknowledgement of cultural and physical diversity is an essential element in a strong and healthy society. I also believe that the conceptualists, designers, and technologists responsible for developing products, programs, and services on behalf of government and industry have a duty to ensure accessibility to the broadest possible range of people. For us this means considering accessibility in the concept design and keeping it simple throughout the development process.

For most people talking bank machines are a new and important step forward in the quest for independence in an ever-changing social and economic system. For others talking bank machines represent a significant advancement toward a future in which all technologies and services are designed to be universally accessible, enabling the full participation of all citizens in the democratic process. And for some people talking bank machines are just another example of rampant liberalism, which will surely be the downfall of the entire economic system by driving the cost of doing business to unimaginably high levels.

Whatever your thoughts about talking bank machines, they are becoming a functional reality in the United States of America. On October 1, 1999, a group of companies with local government support and involvement launched the first American talking bank machine. At a press conference on that day, Susan Leal, Treasurer of the City and County of San Francisco, William Wolverton (President and CEO of the San Francisco Federal Credit Union), Diebold Inc., the Credit Union Cooperative, and I announced the event to the world.

The core technology which led to our involvement in creating talking bank machines was initially developed in 1993 when T-Base was approached by Digital Equipment to provide technical expertise in a joint bid to develop an interactive smart card system for deploying federal employment insurance benefits to Canadians via an electronic network capable of interacting with the existing banking system.

The proof of concept for this project required the development of three software programs: host system software to manage the service delivery, ATM software to manage the human-machine interface, and software to manage smart card interaction with the system. The smart card software was easy. The host and ATM software development proved to be more difficult due to the very limited band width available at that time.

Our lead technical authority, Dan L'Ecuyer, and I developed the plan, and work commenced. Three months later a functioning proof-of-concept host system and ATM were displayed at GTECH (the Government Technology Show in Ottawa, Canada). Unfortunately, the project was shelved due to pending amendments to Canadian Legislation to permit electronic payment of benefits.

However, we used the core of this software with subsequent enhancements to create several other accessible service-delivery systems, namely:

* InfoTouch--the first publicly accessible automated information-access and delivery system capable of producing information products on demand in multiple formats: Braille, large print, audio cassettes, and computer diskettes;

* Universally accessible smart-card system--a cost-effective access module capable of securely interacting with smart cards for service delivery over the Internet; and

* AccessAbill--a system which receives monthly billing and statement data from clients, parses it, translates it on the fly into appropriate formats (Braille, large print, and ASCII text), and prepares for publishing.

In the fall of 1996 T-Base was asked by the Royal Bank of Canada to work as subcontractors to NCR Canada (the ATM equipment supplier) in making the bank's automated bank machines accessible to people with disabilities. We designed the human-machine interface, created bilingual voice files required for audio enhancement, and managed the consumer- and service-provider consultation process. In October 1997 the world's first talking bank machine was launched at the Royal Bank of Canada in Ottawa.

In mid-June 1999 Ms. Laura Arriola, Special Assistant to the Treasurer of the City and County of San Francisco, asked if T-Base could assist the City with the creation of a Talking ATM. An oral agreement with the City of San Francisco Federal Credit Union got us started, and in September, 1999, work began.

Because the base platform is a Diebold ATM running on the Deluxe Network, we requested and received generous assistance from both Deluxe Network and Diebold, Inc., in our development activities.

The first step in the development process was to make the ATM talk by upgrading the existing Diebold ATM with the voice-guidance option as well as scripting and creating the required WAV files, the results being the Talking ATM currently serving the public in San Francisco's City Hall.

In the next phase of our development activities we want to use our core technologies to create an intelligent software module that will be capable of receiving data from the host system in its existing form, parse the data, determine voice file requirements, manage the system key mapping to accept function key or keypad input to provide the user with audio instructions, and return the appropriate data to the host to fulfill transaction requests.

This intelligent software module will enhance rather than replace the existing terminal software. The goal is to avoid the necessity of altering the software currently running on the host system while overcoming local terminal limitations for the provision of accessible services.

Our long-term goal is to provide enough intelligence in the ATM to off-load application code from the host to the terminal. The short-term approach addresses only the issues involving user interaction with the software application. It fails to address a number of issues involving user interaction with the peripheral hardware (for example, cash dispenser or envelope depositor). Context-sensitive voice assistance and smarter handling of error conditions, such as timeouts, are needed. For the ATM to become truly accessible, it is necessary to offer the user much more control over the interactive experience than is currently available. This goal cannot be achieved within the existing host-centric design since it is not practical to maintain remotely a sophisticated human-machine interface.

An additional, and as yet unstated, benefit of addressing accessibility issues in this way is the improved and expanded services which can be made available to all. More intelligence in the ATM will permit a much richer environment to be presented and enable many more services to be developed faster and more cost-effectively than is currently possible, with less lead time to market.

We have outlined our vision for this technology in our paper: "Reversing the Trend: Designing for Accessibility in the Twenty-first Century." It outlines some of the reasons for the failure to consider accessibility issues in the deployment of today's technology-based systems and explains what can be done about it. I invite you to contact me for a copy.

I also invite you, as consumers and providers of services, to give us your comments on the technologies we create and the services we deliver.

Note: Mr. Fowler's e-mail address is <[email protected]>.