Price: $179 with headset; $40 (upgrade); Educational discount available.
Requirements: Mac OS X 10.3.9. Mac with G4 or faster. Universal.
Recommended: Mac OS X 10.4, 512 MB of RAM.
I have been interested in voice recognition software for a while but never played with it. When the chance to explore iListen’s new upgrade came up, I jumped at it. I hit some snags, but I’m pleased to say the product itself is good.
There are two pieces to the iListen setup experience: one is the usual software download or CD loading of the software. No problems there. The other part of the game is making sure you have an approved microphone. I did not anticipate this issue, and it delayed my using the software for several days. If you purchase the software directly from MacSpeech, they include an approved microphone. Since my copy was not purchased this way, I had to scramble for a microphone. Do not assume the microphone or headset you already have will work—MacSpeech has only approved a limited number of units that will work well. Their instructions say that it is possible other microphones will work, but they will not promise excellent results, and they will not provide support.
Once you have the software installed and the proper microphone plugged in, you read a story. Yep. You read a story to iListen, so it can hear what you sound like and how you pronounce words. You will go through the document and make corrections, and read another story or two. Or half a dozen. It depends on how good your accuracy rate is. The FAQ page suggests that once that rate is in the 92% neighborhood, you won’t improve things much by reading any more stories.
My first story came in at 88%. I read two training panels and was using an unapproved microphone—but it is a relatively high-end model, the Plantronics GameCom Pro 1. After one more training panel, my accuracy rate was 90%. I tried to shush the cat during training, but then decided he would be commenting other times, so I might as well let iListen include him in its analysis.
I enjoyed some of iListen’s attempts at understanding the test story I was reading, about new North Carolina laws banning cell phone use by young drivers, and requiring rear-seat passengers to wear seat belts. My favorite error was when I read “Also becoming law Dec. 1 is a requirement for all back seat passengers riding in North Carolina to buckle up.” The software said I was requiring folks to “Bob aloft,” which I thought would be more fun, if you had a hot-air balloon.
I followed the suggestion on the FAQ page, and preset the microphone in the Sound pane of System Preferences. I did not look at the screen while dictating, as that messes with your head on the delay. I used SimpleText, but I would have turned off Spelling & Grammar in Microsoft Word. I turn it off when I use it anyway, and I don’t use Word by choice.
The software works by translating your spoken words into text. It starts with a profile based on your gender and the language you speak. I suppose men with high voices and women with low voices might need more tweaking, if the software expects certain inflections. My voice is low for a woman, so that may have influenced some of my misunderstandings. The requirements do say the user should be at least age 14, so possibly the software is not as good at dealing with children’s voices as it will be someday.
After my fourth training story, my accuracy rate on the driving laws story did not improve beyond 90%. Most of us have to read behind ourselves when we write, anyway. I would prefer not to have to make a correction in every paragraph, but until the software learns my voice better, I guess that’s how it works. It is still miles ahead of where we were a few years ago. Half a dozen training stories at a pop is about all I recommend, as your voice tires. Some of the training stories are a good bit longer, so pay attention to your throat. Some of the errors between my third and fourth training stories changed, making me think I had gone too long for the software to understand me.
The microphone is to be positioned very near your mouth, so it is very sensitive to your voice but not to other sounds. I experimented with having someone talk as he walked behind me while I was dictating. The microphone did pick up his voice when he was within two feet, but not much beyond that. iListen only listens to me! I may set up a profile for the other person, though. He set type for many years but his typing is iffy. Maybe it was because he worked upside down and backwards all those years. He might like to speak aloud his e-mails.
I like this program very much. It is fun to see that we are living in the world our science fiction writers dreamed for us decades ago. I will use this product on my next research paper for graduate school. I have already begun it. The course is Philosophy of Mind; I am eager to see if using this tool changes the way I write or think.
If you spend a lot of time writing, you should consider this product. It is a smidge pricy for my taste, but we get what we pay for, and there is a reason some stuff costs a lot. I could see it being helpful for someone writing a sermon or speech, who thinks best aloud rather than in print. Using this software would allow such a thinker to speak an idea rather than trying to construct it in letters. If you are a walk-around thinker, though, you will need a long cord, as iListen does not yet support any Bluetooth microphones.
I am not sure this software would be a great idea in an office or other environment with several people talking. You would need to test it with an approved microphone to know whether the background voices would be a problem.
The folks this software would be best for, though, are those who have typing issues. I have a wonderful friend with occasional finger problems due to a medical condition, and if he does not already have software like this, I will encourage him to explore it.