UK: VOICE is Just Another Channel of Communication

"Over half of Government services are online; over a million Britons now subscribe to broadband; more than 6,000 UK online centres have been opened nation-wide; and independent research shows that the UK is considered the second best environment in the world for e-commerce. Our goal is to ensure that everyone who wants it has access to the internet by 2005. We are on track for achieving this target. Currently 45% UK households are online and 47% of UK adults are regular internet users. But take-up among the most disadvantaged groups in society – those on low incomes, the elderly and people with disabilities – is significantly lower.

These groups are traditionally heavy users of public services and potentially have most to gain from convenient, customer-focused channels of electronic delivery. Services like benefit applications, access to health records and GP appointment bookings will all be available online. But without access to the internet or the skills to use it confidently, these groups may face further social exclusion. We need to transform digital opportunities available to all UK citizens".

So, how do we transform digital opportunities? We use the most common form of communication?

While the Government understands the issues facing the most needy in society, it is less certain that the steps they are taking will provide the complete answer to delivering eGovernment.

Current thinking and publicity, as directed both to local authorities and the general public, seems to equate "e-government" largely with the internet, although other channels, particularly iDTV, Kiosks and Hand-Helds are also considered. This strategy largely ignores other proven electronic and radio channels available for providing automated services – such as telephone (speech and DTMF), fax, SMS, pager, and email.

The internet is usually the most comprehensive channel in terms of its content flexibility but it is not the most universally available, or the channel of choice in many situations.

Initiatives need to cover more than offering a bigger and better network. Although broadband is highly desirable, or even essential for quick and efficient internet use, its take-up is currently limited by cost, need and availability. In rural areas Broadband availability will be a long-term problem - and there are many predominantly rural local authorities.

It is also not just about creating awareness, or providing training, essential though these steps undoubtedly are. In addition, much more attention needs to be focussed on those groups who are unlikely to ever use the internet, certainly to access Government or Local Government web sites.

The following extract is from the National Statistics report No 32 on Social Trends issued in 2002 and shows the level to which people in Britain are using the Internet.

It can be seen that just 15% of men and 7% of women over 65 are users of the Internet. In fact this figure is inflated, as it includes people who use the Internet for email, but without using the Web. Truer figures for Internet access are closer to 10% and 5% respectively.

One key factor working in the government’s favour is that the "baby boomers", probably the oldest group in society to have a strong grasp of the use and value of IT and the Internet are all too quickly moving towards retirement.

However, the retiring "baby boomers" will bring with them much higher expectations on the levels of service they expect from their local authorities.

Once the investment in technology, education and awareness begins to take effect there will clearly be an increase in the take up in access to electronic services, although not necessarily to the required level.

For those who do have access, there is a new problem – location. Mostly, the Internet is typically used either in the office or at home - much like voice communication used to be conducted only a few years ago. This leads to the fundamental difference between fixed and mobile Internet: services based on all of the three most important value-adding elements of information - personal relevance, location, and time – can only be offered by mobile Internet.

Another study in the National Statistics Social Trends report identified that over 96% of people in Britain use the telephone, while in a separate report 71% of citizens stated that if they were to contact their local authority they would do so by telephone.

It therefore makes excellent sense to use the telephone as the primary access channel to a local authority’s eGovernment services.

For many types of transaction the telephone can be far quicker, more convenient and simpler to use than the internet. It is often the channel of choice - even when internet is also available.

Interactive Voice Response (IVR) systems have been available for over ten years. Most employ a touch-tone DTMF or dualtone multifrequency user interface and deliver a high level of service, particularly if they have been well designed to include "barge-in", short user prompts and limited levels (usually three) of menu options.

As a first step in utilising Voice technology IVR is an excellent option, but it does have its limitations, particularly in the area of speaker verification. IVR often requires lengthy messages, resulting in extended call times.

Another approach is to use Natural Speech Recognition. This is not entirely new, but it is only recently that the technology has matured where systems can reliably recognise spoken responses over 95% of the time; even higher with simple tasks. Key vendors such as Nuance and SpeechWorks are leading the way in offering

the market reliable speech recognition software. Both report that user satisfaction with speech recognition solutions is high, and growing – currently around 90%.

Also featuring strongly in the market are Microsoft® who are offering a .NET Speech platform to make it easier to develop and deploy speech enabled applications.

Why would a local authority consider Speech Recognition as a channel to access their eGovernment services? The most important reason is that it offers extremely easy to use and secure access to the information and services the local authority wishes to make available to ALL of its citizens. Telephone access is the only channel that comes close to supporting this level of service.

Speech Recognition also offers a mobile solution. Users do not need to be seated by their computer, or even in their home or office to access the services offered by the local authority. A mobile phone can be used just as well, and offers opportunities for using services such as messaging and pictures.

Where are the areas of eGovernment where Speech Recognition can add real value? First let’s consider the role that speech is already beginning to play. It does not yet use Speech Recognition in a production service, but Fujitsu’s award wining Avanti solution is using Speech technology to provide real-time advice and guidance to citizens.

authorities one-stop-shop.

The next step is to offer Natural Speech Recognition on input, and when in place the service will be capable of being used by callers over the telephone.

What are the areas of service best suited to a Natural Speech application? We can consider queries such as "Do I need planning permission for my new house extension?", "Will my mother have to sell her home to pay for residential care?", or "What are my housing benefit entitlements?"

For full Speech Recognition the most appropriate applications are probably those that are both simple and repetitive. Typical of a suitable application is a request for a service or information. Please send me a form to ….." Please tell me how I can do this ….."

A common request for a Local Authority is for an appointment. Returning to our two types of user, the disadvantaged may want an appointment with a Social Worker, or to arrange Housing Repairs, while the younger, more mobile citizen may use the service to book an appointment with a Planning Officer, or in the LA’s leisure department, for a squash court.

Other services could include:

• Information services (locations, opening hours, FAQs, etc)
• "Out of hours" emergency calls – highways, engineers, environmental health, planning, housing, etc
• Emergency outbound messaging for engineers, fire, police, ambulance services
• Workforce reporting and monitoring
• Debit/credit card payments for rent, rates, planning applications, parking fines, etc

Some of these may appear to be very simple, but Fujitsu’s experience in Speech solutions for clients shows that the development work should not be underestimated, particularly in more complex automated transactions such as accurately capturing address information. The benefits, however, can be enormous

It would be a mistake to think that Speech Recognition is simply an add-on to existing services. In some cases it may prove to be the case, but it is more likely that the IT department will have to consider the design of the application, particularly the user interface.

Some of the most important factors to take into account include:

Usability. It is essential to user test the new application throughout the implementation. What seems a sensible solution to a young graduate IT development "techie" will almost certainly not be right for a 70 year old, or someone who does not regard English as their first language. Keep the target user base involved.

Credibility. To begin with it may be an uphill struggle to win universal acceptance. All the more reason to get an application in, running, working and seen to be delivering benefit. The more complex areas can follow – as all parties become accustomed to the capabilities and benefits of Speech Recognition.

Personalisation. Try and give the "person" at the other end of the phone an identity. This appears friendlier and helps make it easier for the caller to relate to the service. Also, this makes it more likely the caller will return to use the service again.

Confirmation. Use at different stages throughout the conversation. These techniques help the system understand what the caller wants, while at the same time gives the caller confidence that their message is being understood.

Problems & Queries. Despite everyone’s best efforts the system will occasionally have difficulty with a caller’s voice or message. There will then be a need to ask the caller to repeat the message.

Ensure that the system can handle these issues effectively, but most importantly politely.

Security. This is not an option. It is essential that any application has the appropriate security features, and then is seen to have these features. Callers will be quickly deterred from using the service if they perceive it to be weak in managing the security aspects of their call.

Should a Local Authority adopt a Voice Recognition strategy, what are the typical services their applications could deliver?

Natural Language Recognition, which can include many European & Asian languages

Barge-In – the ability to interrupt the speech engine and continue the conversation

Security – A unique human speech print that distinguishes individual users can be used for secure transactions.

Customer Service / Delivering eGovernment. This is clearly the main driver for a Speech enabled service, and it will make electronic service delivery actually, rather than theoretically, available to all citizens.

Mobile service delivery – it will not only be the disadvantaged who will benefit, but those citizens wanting mobile service delivery.

Improved Local Authority productivity – an automated Speech Recognition system will take pressure off an authority’s Contact Centre for call routing, leading to productivity improvements.

24x7 Service – based on Speech Recognition will help a Local Authority deliver a real 24x7 service

Natural Speech Recognition technology has finally come of age, and is beginning to offer exceptional reliability.

Local authorities wanting to provide real electronic service delivery to all of their citizens, not just those with the ability and inclination to use internet services, would be well advised to consider Speech Recognition as an integral part of their Customer Access Strategy.