|Topic:||Mobile voice in India|
|Organisation:||Telesoft Technologies Ltd|
Bruce Markham is the CEO of Telesoft Technologies Ltd. He co-founded Telesoft, initially as a consulting firm to fund the development of low-cost PC-based telecommunications solutions. They have since developed a wide range of media and signalling solutions for the industry. Mr Markham is still directly involved in product strategy and engineering implementation for both their switching and monitoring product areas. Mr Markham began his career in the telecommunications industry with Plessey specialising in hardware and software development for telephony networks. Bruce Markham graduated from Cork University with a degree in Computer Science.
The interactive voice response (IVR) voice market in India is quite complex, but opportunities are plentiful. Affordable services and a growing economy have driven the growth of mobile phone use. Almost all service there is 2G but given the high illiteracy rates, voice communication is the most effective way to reach the population. India has 21 official languages, but IVRs that recognise ten of these languages can communicate with, and provide services for, 90 per cent of the population.
A growing complex voice market India’s telecommunication market continues to boom with nine million new mobile subscribers being added every month. It has surpassed the USA as the second largest mobile market with 270 million mobile subscribers as of May 2008. However, with a population of 1.15 billion people and an average tele-density of just one-third that of the world average, the market is still far from saturated and continues to grow rapidly. It is estimated that by 2012 there will be 625 million mobile subscribers. The reasons behind this growth are twofold. Firstly, as a vibrant economy its appetite for efficient business-to-business communication links is insatiable, especially in industrial heartlands like Delhi and Mumbai. Secondly, people in India are leading busier lives than ever and using their mobiles to stay in touch with family and friends. Affordable pre-paid mobile tariffs and higher salaries have also played their part in the nationwide adoption of the mobile phone. Verbal communication is particularly important in India, as literacy rates in many rural areas are still relatively low. The 1991 Census (the latest available) showed the huge diversity of languages spoken in India. Whilst English is the natural language of political and commercial communication, there are 21 other official languages! Of these, Hindi is spoken by 40 per cent of the population. Other key languages include Bengali, Telugu, Marathi, Tamil, Gujarati, Kannada, Malayalam, Oriya and Punjabi. Together, these ten languages can reach and interact with 90 per cent of the population. In order for Telecom Service Providers (TSPs) in India to have truly national coverage and support they must cater for at least this subset of ten national languages in order to maintain contact with their customers. The relatively low technology 2G handset has been India’s saviour. Whilst such phones cannot support services that rely on WAP, GPRS or any type of client software within the handset, they do offer verbal communication throughout the Indian sub-continent. Many of the new subscribers use the pre-paid mobile phone service and mobile top-up applications to ‘refill’ their phones with user minutes allowing them total flexibility and control. Opportunities abound The best Indian SPs have turned these twin challenges of language and technology into opportunities and differentiators by introducing voice services valued by their customers. Such services are accessed by those in cities and countryside alike, offering everyone access whether they have high or low literacy skills and in whatever language they prefer to speak. Traditional media platforms have been used for years for pre-paid mobile customer care. Here customers ring in to an automated system and add user minutes to their mobile phone through a pre-payment system. Whilst dual-tone multi-frequency (DTMF) signalling inputs from handset button presses were originally used to input data, voice recognition systems are now improving to the point that users merely speak their wishes in their local language or dialect, and the system understands them and responds accordingly. Automated speech recognition and response systems offer the opportunity to provide many speech-based information services to mobile users in India, such as news and weather services, train and plane timetables, prayer lines, horoscopes, contests and quizzes. The range of services offered by modern media platforms is truly staggering. Today’s media platforms The best interactive voice response systems (IVRs) and their modern equivalent, media platforms, have a number of common characteristics to support these complex requirements. Controlled by IN or VoiceXML server – The majority of telecom networks currently in use still have very traditional architectures, built around the Intelligent Network (IN) infrastructure (CAMEL & WIN). Media platforms must fit into this system as intelligent peripherals controlled from a service control point (SCP) within the network. To fit next generation networks however, media platforms must also connect to IP networks and be controllable via SIP & VoiceXML/CCXML applications. VoiceXML applications typically reside on an application server connected to the media platform via IP. VoiceXML is an extensible mark-up language that provides a ‘voice browser’ interface for users, similar in concept to the visual interface offered by HTML browsers. It is an open language that is easy to learn and code. Application developers frequently build libraries consisting of snippets of VoiceXML code enabling them to quickly pick and reuse these snippets as required to develop final applications for the Indian market. Being an open standard, it has been adopted by media platform hardware vendors as the application development technology of choice and runs on most media platforms today. This allows easy portability and reduces lifetime maintenance costs. The best media platforms allow simultaneous control from both the IN interface, which is fast and reliable, and from the VoiceXML server, which is flexible and more easily managed. Automated voice recognition – The multi-language/multi-dialect nature of India (see above) has proved a technical hurdle for speech recognition software that is only now being solved with the latest generation of software algorithms and hardware processing power. Today’s IVRs however are able to pick out words and phrases in multiple dialects and languages. Connect to traditional and Next Generation Networks – Media platforms traditionally fit into TDM network architectures via high-speed (STM-1/OC-3) or low speed (E1/T1) links and are controlled via ISUP/ISDN. In practice, most IVRs deployed in India for core telephony networks either have this interface built-in or require an external gateway in order for it to talk to the core network. A next generation network (NGN) on the other hand requires an IP connection controlled via SIP. Most modern media platforms have this interface, but relatively few have both interfaces which eases network architecture issues during deployment to core telephony networks and also allows an easy, low cost, upgrade path to NGN networks. Easy hardware and software integration – Because higher-level software applications are always needed to make media platforms useful, easy integration code is important application developers and systems integrators (Sis). Coupling a media platform to a billing application server for instance, and interfacing with the higher-level software application within would be a typical need. From the application software viewpoint VoiceXML compliancy is key, but so is the ease of interfacing to the IN and/or NGN network. Unfortunately, many countries have their own ISUP/INAP variants and so this must also be taken into account when picking a media platform. Typically, SIs use a variety of hardware and application vendors to assemble a solution. A media platform hardware vendor, a different vendor for speech recognition software and, maybe, a third for the application development software tools. They might even buy some or all of the application software packages and databases they need from other parties, making the final solution a complex ecosystem. Hardware and software vendors that work openly together to ease the interoperability of their components help resolve many of these integration problems up front. Sizing the application – Modern media platforms can handle thousands of simultaneous calls in multiple languages. The number of expected callers, the average length of call and the processing capability of the processor define the size and number of Media Platform boxes required. Given the rapid growth of the Indian mobile market, telecommunications operators and service providers (SPs) can expect huge growth over the coming years. The next generation of IVR/media platforms is rising to the technological challenges of the Indian market, chief of which are efficiently handling multiple languages and straddling both the legacy fixed and mobile networks, and the newer IP-based network architectures typically found in networks today. The best media platforms can be controlled by traditional IN nodes (CAMEL/WIN) or by VoiceXML/CCXML services, or indeed a mixture of both, allowing old applications to be re-used and new applications to be quickly developed. This will allow operators to capture their share of the fast growing market and reduce customer churn rates by offering advanced interactive voice services both to lure and delight customers and to grow revenue.