Automatic Speech Recognition (ASR) technology is making waves around the world. Deb Mukherjee says enterprises should be geared to embrace ASR in developing applications that reduce expenditure and lead to better customer satisfaction and retention

Most of us 'love to talk' and get things done rather than punch buttons and keys, or fill in forms. Looks like the time has come for us to do precisely this and jettison our keyboards in the time to come. Yes, Automatic Speech Recognition (ASR) is making this happen and it's only a matter of time before voice, the most natural of all interfaces, becomes the best available alternative.

However, the benefits of ASR are not limited to merely speaking to one's PC. It can redefine the way business is done across the globe.

ASR systems are making round-the-clock access possible to voice portals, providing information ranging from stock quotes and business news to weather and traffic reports, over telephones. These voice portals use speech recognition technology that enable telephones to be used as a means to provide callers with access to information repositories in much the same way that Web portals do as information hubs for Internet users.

Although ASR systems are making headway, they have limited buy-in use if they are not integrated with traditional interactive voice response (IVR) systems. By integrating ASR with IVR, organisations can transform telephony applications (in conjunction with transaction processing systems) to automate applications such as customer order entry, call centres, help desks, and other customer support functions.

We are already seeing a move in this direction with the front-runners being brokerage houses, financial institutions, banks, retail and healthcare organisations. The highly competitive nature of the retail, brokerage and financial services market is making these sectors embrace speech recognition consumer interfaces for a sustained market presence.

The situation today is that if brokerages and financial firms do not integrate speech recognition capabilities with their systems, they run the risk of loosing a significant number of customers to those that do. Higher call volumes are also making the addition of speech recognition more cost-effective in call centres and other customer service organisations.

The primary benefit that motivates these organisations to embrace speech solutions is the potential for dramatic reduction in operational costs. Speech solutions can enhance the productivity of customer service personnel through partial or complete automation of customer calls. Increased automation frees the customer service agent from many routine administrative tasks and reduces costs related to customer service staffing, as fewer agents are able to serve more customers.

Three types of ASR applications readily come to mind--dictation and transcription, command and control, and telephony applications. Dictation applications include simple dictation into windows for example, for creating and editing documents. A few ASR technology providers have succeeded in developing exclusive dictation applications with customised vocabularies for specialised markets such as medicine and law. These dictation products that we see today are also extremely helpful to individuals with physical disabilities.

Further to the benefits of dictation, ASRs can also help in controlling our desktop using customised speech commands. We may just be able to say 'shut down' or 'restart' and our machine will obey. But beyond these simple activities, in the near future, I envisage widespread adoption of ASR technology in telephony applications that includes call centres, brokerage houses, financial institutions and airline reservations.

The list of companies offering speech recognition technology or products and solutions is fairly exhaustive. A majority of blue chip computer software and hardware vendors and communication companies are investing considerably in speech recognition technology.

Undisputedly, financial justification emerges as a critical factor in evaluating new technology solutions for any business. Currently, implementing ASR solutions in an enterprise is indeed expensive. In fact building a state-of-the-art speech recognition and text-to-speech engine make up a majority of the cost. However, with developments in natural language processing (NLP) techniques and speech engines with high recognition accuracy, we can expect the implementation cost to come down significantly. The cost of implementation varies widely depending on the amount of traffic that the set up is expected to handle. A research report on speech enabling a telephony application suggests that the ROI for a typical ASR solution is a lowly 6 months!

In the post-PC era, as computers tend to become more ubiquitous, ASR would become highly significant, and existing interface options such as keyboard and mice will lose out to ASR as the primary user interface. ASR would make our lives much simpler by helping us in a number of activities that we do from browsing the Internet to sending our mails to placing or retrieving stocks to carrying out banking transactions, from anywhere at anytime. And with the current rate of developments in ASR technology, in the near future, we can very well speak with or control our washing machines and refrigerators using voice commands.

ASR coupled with the combined strengths of telephone and the computer would offer great benefits and rapid return on investment. In the current era of fierce competition and where time is money, enterprises should be geared to embrace speech solutions in developing applications that reduce expenditure and lead to better customer satisfaction and retention.

The author is CTO, Cognizant Technology Solutions. He can be contacted at Debmukherjee@cognizant.com