As a writer who has conducted between 2,500 and 3,000 interviews over the last 15 years, I do a lot of transcribing. A lot. The interviews have ranged from 5 minutes to 3 hours each, dependent upon the subject, the topic and our rapport. It’s a miracle my fingers aren’t all scrunched up or my wrist and arm muscles routinely throbbing in pain. One big relief from all of this has been the use of voice recognition software. By listening back to my interview tapes or digital files while speaking them into my PC via a headset, I have been able to cut down on manual overexertion. Granted, there are many errors that come up during vocal transcription that require me to go back over everything again to correct mistakes, but ultimately it’s made my least favorite activity in the world semi-tolerable. I don’t save time, but I do relieve the fatigue and discomfort that comes with long hours of typing.
Then I discovered that I was using the wrong software. When I bought ViaVoice four years ago, I was not sure which would be better, that or Dragon NaturallySpeaking, and none of my writer peeps really knew the answer either as it was new territory for them. I took a chance with ViaVoice. After six months and countless hours of training, it got to be 70% accurate and could recognize band appelations like Dimmu Borgir and Lacuna Coil and rocker names like Tony Iommi and Hansi Kursch. Pretty cool actually, but it still made a lot of mistakes and had problems with my occasional mumbling or fast speaking. Then again, human beings do, too.
When I bought my new PC at the end of this past summer, I decided to try Dragon NaturallySpeaking 10 Standard, especially as it was on sale at Amazon for $45, shipping included. I should have made the transition far sooner. After one hour of voice training and an additional thirty minutes of it scanning all of my Word documents, DSN 10 is already 80% accurate, actually comprehends most of my mumbling — in fact, when I quickly dictated “supercalifragilisticexpialidocious” and “antidisestablishmentarianism” in rapid succession, it got them on the first try — and has made it more of a joy to transcribe. It’s also learning new words quickly, too. And since I have had a stereo digital recorder since June 2008 (the awesome Olympus DS-40), I have learned that I can play back interviewees who speak slowly and succinctly at faster than normal speed and thus get through many interviews far more easily.
My ultimate wish is that Dragon’s parent company Nuance Communications will develop an affordable voice recognition program that can understand multiple voices — such as the ones used by everyone from Delta to Verizon for their customer service phone lines — so that all I need to do is start my voice file and let Dragon work its magic without me needing to touch the keyboard. At least not until the clean-up process where I’d double-check for mistakes and formatting issues. I hope it happens soon. I’d be able to get more work done while my PC handles the dirty work.
Supercalifragilisticexpialidocious Antidisestablishmentarianism indeed!