Moving beyond vanilla cloud and stand-alone IP PBX services into value-added Smart Voice offerings requires a number of building blocks for everything to work smoothly.  I'm hoping cloud vendors currently delivering these services will think about the best way to "play well" with other using open APIs, so developers can use WebRTC and other programming frameworks to quickly build revenue-generating services.

While not a building block per say, voice quality is a high-level requirement for Smart Voice implementations.  Treating voice as data to be processed before piping it through an applications stream means you need to have the best quality data (voice).  HD voice through broadband and cellular services will become increasingly important as telcos finally (belatedly) figure out that having the best narrowband call quality still means it is narrowband, and less desirable than filtered wideband.

The first building block for Smart Voice services is live call recording.  Data (voice) has to be captured for processing in many cases, be it for simple text-to-speech visual voice mail to Siri personal assistant and HyperVoice indexing.   Each phone call adds another record to a database of sales calls and customer interactions, with a volume of calls over time producing mineable data via voice analytics. 

Speech-to-text is the second building block, since raw sound has to be translated into text for use by any other machine application.  Speech-to-text accuracy is affected by voice quality (Hello, HD voice!) and to a certain amount of customization based upon language and geography.

An optional building block to have available is voice biometrics -- the ability to use voice as a unique identifier.  Voice biometrics enables "Fast pass" access to any voice transaction.  Being utilized by Barclays Wealth, TD Waterhouse, Eastern Bank, Vanguard, Turkcell and T-Mobile among others, using voice as a means of authentication can provide better security and faster authentication.   New Zealand government agencies report shaving 28 seconds per call by using voice biometrics.  Multiply that by call center volume.

Blending voice biometrics into wider voice applications has great potential.  Mobile workers could be easily authenticated to provide greater security for many types of business operations,  with voice biometrics acting as a barrier against social engineering hackers.  UberConference and other new generation conferencing services boast of being able to "identify" participants  through a mixture of caller ID linkages and picture tie-ins .  But a truly "smart" conference service (or any auto attendant, for that matter) would be able to recognize the phrase "Doug Mohney, conference," authenticate me, look up the appropriate room, and drop me in on a bridge without breaking a sweat -- that's Smart Voice.

Please join me at the first Smart Voice conference on Tuesday, January 28, in Miami, Florida. The one day event, collocated with ITEXPO East, will focus on the benefits of Smart Voice services such as call recording/archiving, Hypervoice text-to-speech indexing, voice biometrics and voice analytics. The expo provides attendees with actionable insights from top industry experts and companies on current and future state of the technology.