Props to the recently departed James Doohan for my favorite example of botched speech recognition, with Steve Martin yelling at his answering machine in 'LA Stories' coming in as a close second-place.
PC Magazine published an interesting interview with Drs. Robert Sicconi and David Nahamoo, two prominent players in IBM's Speech Recognition division to discuss the current and potential future state of speech recognition in our society. This conveniently comes off the coattails of IBM's official release of Embedded ViaVoice 4.4 which boasts the ability to "comprehend nuances of spoken English, translate it on the fly, and even create on-the-fly subtitles for foreign-language television systems".
This is a technology that has struggled to gain a firm foothold into the consumer market. It has been around in one way or another for decades now (especially as dictation tools) but the complexity of algorithms required to perform reactive analysis of speech patterns, regional dialects, speed, and grammar have slowed down acceptance of the software.
Dr. Nahamoo suggests that the next big breakthroughs in Speech Recognition will be further enhancements to Call Center software, Person-To-Person communication (Language translation), and improved translation of Closed Captioning. From a global perspective, the on-the-fly translation has vast potential, though I will be interested to see how they handle challenges, like tone and context, which are purely subjective. Then again, if you frequently find yourself looking for bathrooms in multiple foreign countries, then your automated language solution is probably coming soon.
Personally, I'm getting impatient to see marked improvements in software designed for automobiles. This article was published almost 5 years ago and engineers are grappling with virtually the same problems today. Yes, the automobile is a noisy environment, but if Bose can manufacture headphones that filter out airline frequencies, why can't this be applied to the same tones emitted in a common driving environment?
So, hurry up and wait for the latest crop of Speech Recognition solutions, and don't throw away your keyboard anytime soon. This is good news for the burgeoning Carpal Tunnel Remedy industry, but not many others.
Recent Comments