Home « Forum « Author Hangout

Forum: Author Hangout

Voice Recognition

red61544

I was reading a story on Literotica and was distracted by the author's misuse of many words. Somewhere in the second chapter, it hit me - he was using a voice recognition program to write the story. Mistakes like "hour" instead of "our", confusion of "two", "too", and "to", "red" instead of "read" happened too many times. Obviously, the program this author was using did not have a logic test built into it. The best of the voice recognition programs automatically test the logic of words in the context of the sentence; the cheaper ones don't. Has anyone ever tried to dictate a story using voice recognition? If so, how successful were you? I've used Dragon which is considered one of the best, but my success with it was limited. I can't imagine using it successfully for anything more than dictating notes.

docholladay

@red61544

Part of that problem is that the wrong word is not always spotted by grammar checkers or other tools.

From what I have observed in the past a voice recognition program has to be trained to recognize the individual user's voice and accent.

Even writers who don't use speech programs have that problem at times. Fingers hitting the keys faster than they realize maybe.

REP

I read a story with similar errors once. I thought the writer was playing some type of game by deliberately using the wrong homonyms. I never considered a voice recognition system being the cause of the errors. Of course, most writers seem to proof and edit their stories before posting them. So that doesn't seem like a valid reason, unless of course the writer is too lazy to do what he should before posting.

Replies:   Switch Blayde
Switch Blayde

@REP

Of course, most writers seem to proof and edit their stories before posting them. So that doesn't seem like a valid reason, unless of course the writer is too lazy to do what he should before posting.


Or blind

Replies:   docholladay  REP  sejintenej
docholladay

@Switch Blayde

Or blind


Definitely. I once had a friend who was blind and used JAWS a major voice recognition program. That was when I first ran across one of the causes of the problem. He was the one who first told me how much time it took to train the program for his usage. One thing he had to learn to do was spell certain words instead of speaking them when using the program.

REP

@Switch Blayde

That did cross my mind. I would expect such a person to have an editor or assistant to help him prior to posting.

Replies:   docholladay
docholladay

@REP

They can be just as proud and independent as anyone else. Some can probably be even more hard headed about doing for themselves.

Just look around at how many so-called handicapped people insist on doing things themselves instead of asking for help.

Replies:   REP
Ernest Bywater
Updated:

@red61544


I've used Dragon which is considered one of the best, but my success with it was limited.


A lot will depend on which version you purchase, and how much time you spend sorting it out to begin with. 20 years ago I worked on a military base and we had to buy an early copy of Dragon Dictate for a Warrant Officer who had suffered a bad injury to his arm, but was retained in a non-combat role for his expertise. The down side was typing the many reports on the computer made his arm worse. Enter Dragon Dictate, 3 weeks of meticulous training to get it to work with his speech patterns, and he was dictating very complex reports with no errors faster than he used to be able to type.

I can't remember his name, but I do know one of the good authors here on SoL uses a much more recent version of Dragon to write his stories, and they don't have any errors in them. He has severe eye-sight issues and does have an editor give them a final going over before he posts the stories, but the editor says there's very little for him to do on them, it's more a confirmation process.

In both cases the key was the training of the software to work the way the user wants. One important way of getting the training right is to have a very different way of saying the commands to the way you say the same words as a text item. As it was explained to me, when he says the word command as a text item is sound like co-man-d (like most people pronounce the word), while when given as the code word to change to the Command Function mode he says it as coo-ma-end so it sounds very different. The same is true of many other of the function words. This was a deliberate act in the training process to minimise the confusion. He also does something similar with many of the words that sound similar - to=to, to-oo=too, two-be=two type of differences in the words. A lot of memory and training needed, but he gets a perfect result in the output.

typo edit

REP

@docholladay

Yes I know, although my experience around people who are blind or have minimal vision is limited.

Crumbly Writer

@REP

Yes I know, although my experience around people who are blind or have minimal vision is limited.

For one thing, they're more likely to have the text read back to them, which is an excellent way to catch many of that type of error. Like with Siri, the speaking of pronunciation is necessary (i.e. "To be comma or not to be period."). With "to"/"too" and "two" you'd typically spell each one out.

There have been a few authors here who have used T2S (Text-to-Speech) programs over the years, but almost no blind authors (that I've noted).

Ernest Bywater

@REP

Yes I know, although my experience around people who are blind or have minimal vision is limited.


My mother was born with a rare eye disease that saw her go progressively blinder. By the time I was old enough to go to school (youngest of 3 kids) mum was totally blind. Thus I grew up in close contact with a blind person. Mum had learned to touch type while younger and would type her own letter on an old 1930s Remington typewriter. However, sometimes she'd get off line and the letters looked like gibberish, until you retyped them with the fingers offset by one spot.

Mum was very independent and could get around a lot by herself, but she also knew when she had to have help and asked for it.

Replies:   docholladay
docholladay

@Ernest Bywater

Mum was very independent and could get around a lot by herself, but she also knew when she had to have help and asked for it.


Heck we are all like that in some ways. I think that line from one of the Dirty Harry movies almost had it right: "A man has to know his limitations". Should have included the following: "and be willing to admit them by asking for help as needed". Of course its one of those things I learned the hard way.

sejintenej

@Switch Blayde

Of course, most writers seem to proof and edit their stories before posting them. So that doesn't seem like a valid reason, unless of course the writer is too lazy to do what he should before posting.

Or blind

There was a UK paperback author whose books one saw in the airport newsstands (he had a cousin of similar family name who was also an author). I only found out after he had an operation in the early 90's that he had been blind since about 1948 but he was writing his books and reading with software. He admitted that it took a long time to get the software right but conversing in newsgroups his grammar and spelling was logical and correct.
As for avoiding homonyms I am using early 2000's software to check my French grammar and spelling and it is excellent even where my speeeeeeeling is way out.
I would have thought that using a combination a writer could at least double check what is appearing on paper.

As for CW's comment, perhaps it is my accent but, speaking slowly and clearly, I have almost never had problems with SIRI or with Cortana

Replies:   Crumbly Writer
Crumbly Writer
Updated:

@sejintenej


As for CW's comment, perhaps it is my accent but, speaking slowly and clearly, I have almost never had problems with SIRI or with Cortana


From posted reports, the next version of Siri (I forget the new name) will be hands and shoulders better than the current version. Apple bought the company that developed Siri but never hired the designers. They've again purchased a whole new company to acquire their technology, which we won't see until the next iteration of iOS devices.

The key difference, reportedly, is that it'll handle complex questions with 80% efficiency, as opposed to 30% with all the current technologies. There's no telling what it'll look like when it finally appears, though.

Replies:   Ernest Bywater
Ernest Bywater

@Crumbly Writer

There's no telling what it'll look like when it finally appears, though.


sure there is, it'll look like an iPhone!

Replies:   Crumbly Writer
Crumbly Writer

@Ernest Bywater

sure there is, it'll look like an iPhone!

It's funny how things turn around. When Apple first ran their first big ad about 1984, the lamented how everything looked the same, and now, every phone looks similar to the iPhone. Payback is a bitch! 'D

Replies:   Dominions Son
Dominions Son

@Crumbly Writer

When Apple first ran their first big ad about 1984, the lamented how everything looked the same, and now, every phone looks similar to the iPhone. Payback is a bitch! 'D


Form follows function. There aren't many functional options for forms for a touch screen driven smart phone, and none that would look drastically different from an iPhone.

Back to Top