Writer meets arthritis
December 3, 2009 11:18 AM Subscribe
Mac voice-recognition software for a writer with arthritis.
My father's arthritis is making it more and more difficult for him to type. This is hard for him, because he's been a writer for decades (over 25 books and countless articles). He is considering switching to voice-recognition software.
Here are some things to note:
- He is originally from England but has lived in the US since the 1950s. He is a very clear speaker, but his accent is a mix of British (cockney originally) and American.
- He owns a Mac.
- He is not highly computer literate, but he lives in a university town and could find people to help him set things up if necessary. He probably will have trouble if the voice-recognition software itself is overly complicated to use.
I am looking for any advice and experiences with this. What is the state of the art these days? What's available for the Mac? What is the experience like for people who are heavy users?
My father's arthritis is making it more and more difficult for him to type. This is hard for him, because he's been a writer for decades (over 25 books and countless articles). He is considering switching to voice-recognition software.
Here are some things to note:
- He is originally from England but has lived in the US since the 1950s. He is a very clear speaker, but his accent is a mix of British (cockney originally) and American.
- He owns a Mac.
- He is not highly computer literate, but he lives in a university town and could find people to help him set things up if necessary. He probably will have trouble if the voice-recognition software itself is overly complicated to use.
I am looking for any advice and experiences with this. What is the state of the art these days? What's available for the Mac? What is the experience like for people who are heavy users?
Response by poster: Thanks, Tchad. Just to be clear, his primary reason for writing is to write letters, stories and essays. That's basically what he uses the computer for.
posted by grumblebee at 11:38 AM on December 3, 2009
posted by grumblebee at 11:38 AM on December 3, 2009
Voice recognition software has made a lot of strides in recent years, but it still works mainly within a restricted set of voice inputs (i.e. it's best at telephone voice response systems where the speaker will select from a limited number of things to say, precisely the market this technology is designed for). For general dictation, it's still not there.
That's not to say it may not be useful. But I'd also look into adaptive input devices. For example, this wand-operated keyboard designed for the disabled. Possibly you could combine the two. Your father might be able to use a voice recognition program like MacSpeech Dictate (a mac app that uses the Dragon engine, which is probably the most successful PC speech application) to get a first draft down quickly, then go back and fix the inevitable mistakes with the wand keyboard.
posted by Naberius at 11:38 AM on December 3, 2009
That's not to say it may not be useful. But I'd also look into adaptive input devices. For example, this wand-operated keyboard designed for the disabled. Possibly you could combine the two. Your father might be able to use a voice recognition program like MacSpeech Dictate (a mac app that uses the Dragon engine, which is probably the most successful PC speech application) to get a first draft down quickly, then go back and fix the inevitable mistakes with the wand keyboard.
posted by Naberius at 11:38 AM on December 3, 2009
I use MacSpeech some. It is better than other things I've tried.
posted by hydropsyche at 12:42 PM on December 3, 2009
posted by hydropsyche at 12:42 PM on December 3, 2009
For general dictation, it's still not there.
I really have to disagree with this. About 5 years ago, I was forced to rely on Dragon pretty much exclusively for writing. It's certainly viable once you get past the initial learning curve, and I was making my living on the written word at the time.
The big problem is that the learning curve is a doozy, some of the caveats:
1: The accuracy of the Dragon engine improves with context. So you have to give it complete sentences and avoid the urge to treat it like an idiot or small child.
2: It doesn't work well with proofreading and editing on the fly. This arguably isn't a bad thing because a write-proofread-edit workflow is actually better in some respects than on-the-fly editing.
3: It's a bit twitchy in regards to tone and vocal strain. Dictating is actually pretty hard work when you get into it.
4: A good headset is a requirement.
5: It's a mutual-learning process. If you invest the time, you can reach high accuracy and speed, but the first week is frustrating.
The state of the art is probably Dragon, which apparently runs under Parallels and Virtualbox. MacSpeech Dictate is an option now that it has ported Dragon's correction and editing functions.
posted by KirkJobSluder at 1:54 PM on December 3, 2009
I really have to disagree with this. About 5 years ago, I was forced to rely on Dragon pretty much exclusively for writing. It's certainly viable once you get past the initial learning curve, and I was making my living on the written word at the time.
The big problem is that the learning curve is a doozy, some of the caveats:
1: The accuracy of the Dragon engine improves with context. So you have to give it complete sentences and avoid the urge to treat it like an idiot or small child.
2: It doesn't work well with proofreading and editing on the fly. This arguably isn't a bad thing because a write-proofread-edit workflow is actually better in some respects than on-the-fly editing.
3: It's a bit twitchy in regards to tone and vocal strain. Dictating is actually pretty hard work when you get into it.
4: A good headset is a requirement.
5: It's a mutual-learning process. If you invest the time, you can reach high accuracy and speed, but the first week is frustrating.
The state of the art is probably Dragon, which apparently runs under Parallels and Virtualbox. MacSpeech Dictate is an option now that it has ported Dragon's correction and editing functions.
posted by KirkJobSluder at 1:54 PM on December 3, 2009
I have a client that runs Dragon Naturally Speaking under Parallels, it is quite impressive.
I'm interested to try MacSpeech Dictate, but haven't had the chance too yet.
posted by Wild_Eep at 2:21 PM on December 3, 2009
I'm interested to try MacSpeech Dictate, but haven't had the chance too yet.
posted by Wild_Eep at 2:21 PM on December 3, 2009
Or for a shorter answer, IME the Dragon engine works very well for writing, assuming that you are willing to change your habits around its capabilities.
posted by KirkJobSluder at 2:32 PM on December 3, 2009
posted by KirkJobSluder at 2:32 PM on December 3, 2009
So, I'm writing this post meta-filter, specifically using Mac dictate. I've never used Dragon naturally speaking, it's reputation is pretty well known.
You really have to get used to using speech recognition. I find that it's writing from a different space -- just like audio books are different way into my brain than reading is.
I did some minor corrections to this post -- I'm trying to leave the mistakes in my text. I find that I have to pause, collect what I'm about to say, and then say it for the speech recognition to work well.
The training of these tools has gotten progressively faster and easier. If he writes particularly technical work or jargon based information, he'll find that it's best for him to train Mac dictate with some texts of his own.
The biggest frustration I have is the idea that I have to constantly be adding punctuation and editing my text as I go. It's that last one I can be really annoying. Every time that the software makes a mistake I have to correct it immediately (and this really disturbs my flow and the train of thought.)
Software comes with a decent headset, requires approximately an initial hour of training (well actually it can be quite a bit less than that.) Further training really improves the recognition.
My other major frustration is that the software works best in the their interface, their text editor and somewhat less well inside of other applications. I find that I use their sort of Notepad program to do most of my writing (speaking), and then find myself cutting and pasting it into a different document. It has a floating window overlay allowing you to quickly see what it thinks it's guessing. It also has the ability to have a command mode and the spelling mode aside from speech recognition.
posted by filmgeek at 7:43 PM on December 3, 2009
You really have to get used to using speech recognition. I find that it's writing from a different space -- just like audio books are different way into my brain than reading is.
I did some minor corrections to this post -- I'm trying to leave the mistakes in my text. I find that I have to pause, collect what I'm about to say, and then say it for the speech recognition to work well.
The training of these tools has gotten progressively faster and easier. If he writes particularly technical work or jargon based information, he'll find that it's best for him to train Mac dictate with some texts of his own.
The biggest frustration I have is the idea that I have to constantly be adding punctuation and editing my text as I go. It's that last one I can be really annoying. Every time that the software makes a mistake I have to correct it immediately (and this really disturbs my flow and the train of thought.)
Software comes with a decent headset, requires approximately an initial hour of training (well actually it can be quite a bit less than that.) Further training really improves the recognition.
My other major frustration is that the software works best in the their interface, their text editor and somewhat less well inside of other applications. I find that I use their sort of Notepad program to do most of my writing (speaking), and then find myself cutting and pasting it into a different document. It has a floating window overlay allowing you to quickly see what it thinks it's guessing. It also has the ability to have a command mode and the spelling mode aside from speech recognition.
posted by filmgeek at 7:43 PM on December 3, 2009
Coincidentally, Dragon Dictation was released for the iPhone today, with a voice search app called Dragon Search to follow shortly.
posted by fairmettle at 5:12 PM on December 8, 2009
posted by fairmettle at 5:12 PM on December 8, 2009
This thread is closed to new comments.
When I forget and leave it on, sometimes general conversation cues it to do something, but it announces what it is doing, so it isn't the end of the world.
When giving the computer verbal commands, I find that if the setting is just right, it works maybe 85% of the time the way I want it to. When there is a lot of noise or reverb, that goes down to about 15-20%.
I do find that I have to use my clean radio voice for the best result - if I slip into my Ohio River Valley twang it mis-hears me a bit.
I like the ability to have the computer read emails to me, though. It took a minute to get used to the computer voice (I prefer the one named 'Victoria').
That's my experience - I cannot speak to other software, but would suggest getting a mic to speak into. The one built into mine is a little tinny (which could be why it doesn't hear me properly).
posted by Tchad at 11:35 AM on December 3, 2009