The following information about voice recognition and speech recognition was copied with the permission of the owners of the website, www.courtreportingschoolsonline.com.
VOICE RECOGNITION TECHNOLOGY
Surveys of national captioning companies, including some of the largest to the smallest, found that these companies are not employing stenomask writers with voice recognition translation capabilities. Voice recognition has not been successful for realtime writing for broadcast
closed captioning or realtime court reporting. Based on our surveys of closed captioning companies, as well as realtime court reporting firms, our conclusions are that schools and programs advertising voice recognition translation are exaggerating the success of this technology as employed in real-time court reporting and captioning. Polling of court reporting firms and closed captioning firms in one of the largest states concluded not one company was employing voice recognition technology in their business. Successful examples of voice recognition translation in realtime court reporting are extremely rare. Schools claiming 6-month to 1-year completion rates in voice recognition translation software are exagerrated, and we strongly advise the prospective student to canvass closed captioning companies as well as court reporting agencies before investing thousands of dollars and time and effort in this old-fashioned technology. The software has been around for years, but very few people have mastered it sufficiently enough, even over a 20+ year period of time, to be able to offer realtime translation.As of this writing, the National Verbatim Reporters Association has not had one realtime reporting voice recognition translation reporter pass their national realtime certification examination in six years. At least one half of the United States do not allow steno mask or voice recognition in their court systems. The National Court Reporters Association does not recognize voice reporters or voice recognition in their organization. After attending two recent voice writer conventions, demonstrations by two of the best realtime voice recognition reporters in the country were only accurate up to 150 words per minute, far from the necessary 180-200 words per minute required for realtime translation certification. Voice translation software, even with its very limited capabilities, is quite expensive and requires at least a $3,000 to $4,000 investment in the hardware alone, not including the substantial investment in the voice translation software. Generic computers purchased from general office supply stores like Office Depot, Best Buy, Circuit City, etc., will not work with voice translation software, because after about one hour of voice translation, the computer will overheat and crash. They are required to purchase specially built computers costing approximately $5,000 containing additional fans to cool the computer. Voice recognition translation software is used on a limited basis when we call certain businesses prompting us to say yes, no, representative, etc. We can even program names into our cell phone and have the numbers automatically dialed. Some doctors have been able to utilize voice translation software, because they say the same words repeatedly. That software requires the speaker to leave a split-second interval between each word, and no one speaks in that manner. Even these limited applications of voice technology do not always work correctly. You may often say one thing, like Payment History, and be directed to the wrong department, like Sales, which doesn't sound even close to Payment History. You may say Sally and have the phone dial Robert, which sounds nothing like Sally. Certainly when we have numerous attorneys speaking at the same time, perhaps some with a foreign accent, it is impossible for realtime voice writers to designate the speakers and translate all the words accurately.
