Voice Recognition Strives for Success in the Hear and Now


Voice recognition touches on many concerns for IT departments, especially those with large mobile work forces. It is the ultimate user interface and, as such, can make convergence of applications more efficient and drive productivity.


Voice recognition also is a vital safety feature, and adding safety to any sales pitch is a sure winner. That's why Nuance Communication's 2008 In-Car Distraction Study is good news for the category. The test had 30 drivers manually select music on an MP3 player, make calls and program an address into a GPS device. (It doesn't say whether signing insurance and liability waivers also was part of the test.)


The story walks through the statistics. Suffice it to say that voice recognition reduced measured distraction and kept the vehicle in the "ideal car position" a higher percentage of the time. It's intuitive: The more voice recognition is used and the less tasks are done manually, the higher the percentage of time drivers' eyes are where they should be -- on the road.


It's not all about safety, however. Voice recognition as a user interface makes it easier for mobile employees to juggle applications on mobile devices. This bMighty post on working on the road describes Dragon Naturally Speaking. The writer says that it is a "highly evolved speech-to-text" package that works directly in the computer or, for transcribing at a different time then when the words are spoken, through a digital tape recorder.


TechCrunch IT throws some cold water on the feel-good story of voice recognition, however. The writer does this through a close look at Say2Go, a new Windows-based IM client. Bill Gates, he says, predicted 11 years ago that voice recognition would be a common user interface technology by now. It hasn't happened, and the writer points to Say2Go as an example of why the industry segment is struggling. The "training"-- the reading of text into the device to give it a reference point -- is time-consuming. Microsoft aims at 80 percent accuracy, but the writer says that that isn't realistic. Say2Go, he said, made numerous mistakes and lost the meaning of whole sentences.


The category is soldiering on, however. There's a lot of good information in this post, which essentially is a complaint that Google is getting inordinate credit for adding voice recognition to Google Mobile Maps long after Windows Live Search did so last year. In both cases, a voice clip is recorded, sent to Google or Microsoft, and the requested information is transmitted to the handset. It's interesting that Google is testing its system on business-centric BlackBerries.


David Pogue at The New York Times this week reviewed Yahoo's OneSearch With Voice. The service uses speech recognition to decode a request. The terms are run through Yahoo's search engine and the links sent to the mobile device. Pogue says mobile speech recognition generally is limited to very few words (ie, "sports," "weather"). In the case of OneSearch With Voice, however, the software was completely accurate with 17 disparate terms or phrases. It missed on four others. So far, it is available only on the BlackBerry Curve, Pearl and 8800.


The speech and voice recognition sector is fascinating. Creating software that can accurately decode and transcribe speech -- especially when it is received from a handheld operating in a noisy environment and over a wireless network -- is very difficult. However, the profits to those that do succeed ensures that many vendors will keep trying.