Want to Build a World Wide Voiceweb?

In a video-centric industry like ours, it’s tough to get attention for basic voice applications, but I keep trying. A couple of months ago, this column talked about the opportunities that could surface if our industry were to commit to upgrading voice customer premises equipment (CPE) devices (aka telephone sets). This month, we’re going to look at a set of telephony applications that can be sold without doing anything to CPE, which gives a subscriber the feeling of high tech and yet can be used by the technophobic progress-stopper whom we all know and love. In a nutshell, these applications use network-based voice recognition tied to automatic call dialing, routing and some data storage. To the consumer, it could mean the beginning of a voice analogy to the World Wide Web, a new tool that allows consumers to pick up the telephone and talk to it to interact with databases and services. A possible first step One of the possible first steps to a "voice web" is currently being marketed direct to consumers by SpeechPhone LLC and its marketing arm, Next Generation Business Systems (www.ngbr.net), as a product called MANDI (My Automated Natural Dial Interface). MANDI initially targets simple telephony applications. For a find-me service, it works like this: You subscribe to MANDI, which gives you a local access phone number to use as your published phone contact. When I call your MANDI number, I am connected to your automated personal assistant. The automated assistant asks me the name of the called party and my name. It then places me on hold while it autodials your preferred actual reach number and, if you answer, announces my call with the recorded name I have provided. You then decide whether you want to answer the call or request I leave a message. The automated assistant will inform me of your choice. If you’re not available, I’m asked if I want to leave a message or try to find you. If "find you" is the option I choose, the automated assistant places me on hold and successively dials the next most likely phone number you have provided it. If that doesn’t work, the assistant announces failure to find and states it will go to the next number. When all possibilities are exhausted, the assistant asks if I want to leave a message. MANDI also includes caller ID and number storage, so the next time you want to call me, you can simply announce my name, and the system will dial my number via its interconnection to the public switched telephone network (PSTN). Service packages are offered that provide basic MANDI functions and tiers of network usage minutes for a flat monthly fee. MANDI is based upon a network-resident softswitch with an interface to a Lucent voice processing system. The softswitch can be accessed via either a PSTN connection or an Internet protocol (IP) address. SpeechPhone uses Level 3 for local access to the co-located softswitch, which allows it to provide local telephone numbers in more than 3,000 cities across the United States. For voice activated outgoing calls, subscribers can either dial an access number to the system or use a broadband connection with a telephone adapter to receive a voice prompt rather than dialtone when they pick up the telephone handset. New … and not Admittedly, neither voice recognition nor automatic dialing is new technology. However, putting the two together with a user-friendly interface and a marketing program is a novel concept. Once the door is opened with this application, other speech interface applications can also be sold. SpeechPhone CEO and MANDI developer Michael Metcalf states a vision that is worth emulating: "Our goal is to make speech to the phone as a browser is to the Internet." Going beyond the basic find-me and autodial applications does require some basic computer skills, but they’re well within the proficiency level of broadband data subscribers. MANDI includes the capability for computer-based dialing and downloadable PC icons to activate additional applications. Metcalf explains the implications: "With a link to Plaxo.com, for example, we have the capability to download your contact list into the voice messaging system, which opens up a number of possibilities. Obviously, you will be able to call any number in your address book by simply speaking the called party’s name, but you can also ask MANDI to send an email, which is a transcript of your voice recording, to that person’s email address." Other applications include simple audio database retrieval, such as the "Bible by Phone" offered by one of Speechphone’s clients that reads chapter and verse upon spoken request. Not an endorsement Lest I be misinterpreted, this discussion is not an endorsement for MANDI or any other product. MANDI is an example of application implementation that opens the door to more revenue from voice services. As I said earlier, neither voice recognition nor automatic dialing is new, and the MANDI architecture is not the only way to tie them to a telephony service. The type of applications in this column also seem to be an ideal fit for integration of those proven technologies into a PacketCable Multimedia (PCMM) architecture, with a speech processing system accessed via an applications manager. IP multimedia subsystem (IMS) technology is already being incorporated into PacketCable. That includes the Home Subscriber Server (HSS), which will keep track of where a subscriber is most likely to be found, giving alternate options for routing a call. However, it will be a business case analysis that determines whether a speech browser is good business, and if so, whether the most appropriate path to it is via PacketCable, or via a business relationship that allows a cable operator to market products like MANDI, or by firms such as Speechphone merely using cable’s broadband transport in a similar manner to some voice over IP (VoIP) offerings today. Both IMS and PacketCable are complex systems integration exercises, and it will take some time to trial and commercially introduce features. It may make sense to begin offering a solution that is available now to get feedback on consumer acceptance and behavior, especially if there’s nothing at the customer premises to change when we upgrade our networks. While we’re at it, we might just make a couple of dollars and position ourselves with an offering that further differentiates us from the telephony incumbents. Justin J. Junkus is president of KnowledgeLink and telephony editor for Communications Technology. Reach him at [email protected].