Talk to Me

Speech is everywhere.  By everywhere, I mean everywhere in the world of technology and interaction. And it is getting really good.  Even IBM has seen the light (IBM – Watson to Super Siri) Some may argue, however, I would posit that the more significant problem is the implementation of speech and not the technology itself.  But more on that later…

My first work with speech recognition was many years ago, working for a couple of very advanced cardiology firms in Kansas City.  At the time, we were experimenting with electronic medical records.  We were also exploring how the cardiologists could dictate chart notes and have them automatically transcribed into the patient record.  At the time, Dragon, then a very small firm, had an application that could be loaded on a machine and in only two weeks, learn the style and tone of the person dictating.  In case anyone was curious, doctors as much as i appreciate them, have zero patience for such things. A couple of adventurous docs worked with us, but in the end we passed on any sort of deployment.

Times and technology have changed.  I’ve gone from technology that required 2 weeks of constant training (with questionable accuracy) to capabilities that can sit on your mobile device, recognize your voice through cloud services (thanks Nuance), and provide a complete multimodal experience.  All in less than 2 seconds.  Speech is in your car, built into your computer, and even into your television.  While hopefully more friendly than HAL, this era of speech enablement shows incredible promise.  It will take some time to get people acclimated to a new normal. As long as there is a focus on quality (take note auto manufacturers – we can’t have a static platform – you need to embrace the cloud) and a strong element of design (the right use of speech and visual prompts), speech will become a natural way for us to interact – with technology.

Welcome to a new era of interaction – talk to me.