Warning: your call may be recorded for training and speech recognition analysis purposes from a cloud-based software service
So, you’re setting up a call centre IT backbone in your role as IT manager or consultant and you start by drawing up a procurement list. Desktops, headsets, screens and half-way intelligent phone units are all there along with essential back office provisioning including servers (physical and virtual) with extended cloud flexibility options for peak periods of data storage and client data processing.
Your planning is completed right down to Post-It notes and pencils - so that should just about do it right? Hang on, what about CRM-centric speech recognition technology delivered via a cloud-driven Software-as-a-Service model? There’s no need surely - after all what do we employ phone “reps” for anyway if not to be able to “recognise speech” down the phone line when they hear it?
Ah but what about those pre-connection warnings that tell us that our call “may be recorded for training purposes”, what exactly do these companies do with our call data?
OK yes, if you swear down the phone as you F-and-blind your way through a complaint then you’re probably not going to get analysed too much; and many calls will need a real human operator to “listen back” and try and work out a resolution to customer’s complaints or requests.
But the fact is that there is a whole other “glut” of speech data just waiting to be analysed by Customer Relationship Management (CRM) software if these voice records are to be “leveraged” to their full benefit.
This is a key area for companies like Nuance with its specialism in so-called Natural Language Understanding (NLU) technologies that work at both the cloud-centric enterprise back end, as well as the consumer front end, with its arguably fairly popular Dragon NaturallySpeaking product.
In fact, the Dragon product is responsible for extended usage of cloud hosted facilities as the free-of-charge iPad version has been around for more than a couple of years now. “As tablets start to use more sophisticated dictation technology such as Dragon Dictation Free, we clearly get to a situation where the devices themselves do not have the storage capabilities we need to keep all this data, so it’s a natural fit for the cloud,” says Peter Mahoney, senior VP and chief marketing officer of Boston-based Nuance.
Mahoney’s corporate line on how Nuance monetises the free iPad app is somewhat tight-lipped; the firm clearly makes its money elsewhere and is happy to fuel the channel for voice data storage in the cloud. This of course “organically” improves the collective analysis engines that this technology needs to be able to interpret the nuances of different languages, dialects, tones and the multifarious idiomatic idiosyncrasies that exist across the global speech spectrum.
So it all starts to make a bit more sense now. Nuance bought voice-to-text transcription company Spinvox back in 2009 and so has been using cloud-based resources to refine its Natural Language Understanding databases, which logically work by contextual-interpretation in order to distinguish between homonyms (words that sound the same but have different meanings) such as a ‘their’, ‘there’ and ‘they’re’ for example.
Nuance’s technology chops up speech sound inputs into digital signals which are then further dissected into component speech sounds known as “phonemes”. The upshot of all this should (I hope) sounds less of a ‘curve ball’ that it did at the start of this story i.e. the cloud is a good fit for speech recognition data that is being used for call centre CRM-style customer conversation analysis - and the total database of speech data is further fuelled by tablet-based speech recognition apps powered from the cloud as well.
Speech recognition technology has natural home in the cloud then does it? That wasn’t a surprise after all right?
No comments:
Post a Comment