Mondegreen
Last year a new word was added to the English lexicon mondegreen. Wikipedia defines it as: "the mishearing or misinterpretation of a phrase, typically a standardized phrase such as a line in a poem or a lyric in a song, due to near homophony, in a way that yields a new meaning to the phrase."
Here are some examples in song lyrics:
"There's a bathroom on the right" (the line at the end of each verse of "Bad Moon Rising" by Creedence Clearwater Revival: "There's a bad moon on the rise")
"The ants are my friends " (from a lyric in the song "Blow' n in the Wind" , by Bob Dylan: "'The answer my friends").
Voice Dictation Finally Gets It Right
When you consider how easy it is for the average person to misinterpret what they hear and inadvertently give it a whole new meaning, just think of how extremely challenging it is for a machine to capture what you say and accurately convert it into a written word or phrase.
IBM's Thomas J. Watson Research Center began developing computer speech technology as early as the 1960's. I have long been a fan of voice dictation. I used the early version of IBM's product, ViaVoice, back in the 1990s when it was still being perfected. As personal computers became more powerful, voice dictation started to migrate from room-sized mainframes to desktop computers and eventually to laptops. Early PC-based voice dictation systems were commonly referred to as processor "hogs" due to the fact that they would burden any Intel 386/486-class processor to the point of slowing the computer to a crawl. They also contributed to many appearances of Microsoft Window's "blue screen of death" computer crash.
We've Come a Long Way
Modern voice dictation systems like Dragon Software's NaturallySpeaking 10.0 are finally delivering on the promise of accurate and responsive voice dictation.
If you're like me and dread typing multi-page documents or you can't even type the word "nirvana" wrong so that Microsoft Word will give you a choice to correct it with the right spelling, just saying "nirvana" into NaturallySpeaking 10.0 and having it magically appear in your sentence with its correct spelling is like, well, nirvana!
So here I am writing this blog posting with my voice. How accurate is it? Based on personal experience, it's at least 95% accurate. It all depends on your processor speed, the amount of memory you have available, and how much training of the system you've done. Training the system is accomplished by reading sample stories aloud during the software setup process and later as needed to increase accuracy.
Dragon NaturallySpeaking 10.0 and IBM ViaVoice are published by Nuance, a company located in the Boston area. Their products are currently available only on the PC platform, but there are other companies making versions for the Mac. A new laptop notebook or netbook is more than capable of giving you responsive and accurate voice typing capability.
What I Love About It
My current choice is Dragon NaturallySpeaking 10.0 Preferred Edition which includes a Plantronics Bluetooth wireless headset and adapter for use with a laptop computer. The wireless headset frees you from the uncomfortable and weighty wired versions that can often get in your way especially when using a laptop. You can also use this Bluetooth headset your with cell phone when you're not using it for voice dictation.
One major additional benefit with voice dictation software is the capability of most voice dictation software to read back aloud, albeit in a computer voice, what you have typed or dictated. This is extremely useful when writing anything that may contain words that look closely alike but have completely different meanings. For example, thanks to voice dictation software that proofread aloud some advertising copy, I was able to catch what could have been an embarrassing mistake interchanging soup "bowels" for soup bowls.
Cut Me Some Slack, and I'll Give You Some Time Back
We know that mishearing and misinterpreting spoken words is common among humans and is now even defined by a new word. Given that, isn't it time to have realistic expectations of the capabilities of a computer listening to your voice and typing words? Keeping that in mind, you can use voice dictation very effectively to "type," that is, if you don't mind a little real typing and editing to go along with the speaking. When all is said and done, using your voice to type translates into productivity that can help you get some of your precious time back.
I hasten to add to your eloquent diatribe above... that on the Apple Mac platform there is an even more accurate and elegant piece of dictation software called --> Dictate! The software has been around since Dragon's early days but has always been in the shadows of its PC brother. It is built off of the Phillips speech algorithm and accepts a variety of input devices - making it very versatile for both field and office work.
ReplyDeleteBut it is now out there... I am using it to write this piece back to you.
Cheers - Peter
Rick, Nuance is (or at least was) the recognition engine behind many of the "big" systems used by companies to update their IVR systems. The accuracy of those systems is, IMO, amazing. You run across it all the time - phone companies use it, Microsoft uses it for activating Windows.
ReplyDeleteBut 95% is still a problem! Did you find the typos in this post? There's more than one... Therein lies the problem, the errors are often reasonable, which actually makes them harder to find. OCR used to boast about 99% accuracy, which still meant 20 errors on a typical page.