"Wes Groleau" <groleau+news@freeshell.org> wrote in message news:yz9rj.198$qw4.29@trnddc02... > Kerry Raymond wrote: >> The problem is the little words ("and", "are", "at", "or", "on", "in", >> "the", "they", "there", "their" etc). When we speak a sentence normally, >> we tend to say these little words quickly and often only make the vowel >> sound ("ah", "eh", "oh" etc) which makes it very hard to recognise >> exactly which > > I haven't noticed this problem in either ViaVoice or Dragon Naturally > Speaking. Both do some sort of statistical analysis that uses nearby > words to help decide. Agreed. I also noticed a dramatic improvement when I upgraded to the versions above 7. It still can't tell the difference between words the same (eg knight and night) but it would be unreasonable to expect it to do so. Proof reading and some correction will always be needed, but I estimate that it's 95% accurate, or more. Since I'm from one of the generatins when they thought that academically-inclined children didn't need to learn to type, it's MUCH better than anything I can achieve. > > The problem I HAVE noticed is that I need to keep looking at the text and > avoid looking at the screen. Any time I pause, the program tries > to transcribe every tiny little sound that happens as one of those > "little words." Yes. Since I bought a better microphone, I no longer have trouble with background noise such as traffic or people walking past my office, but it's interesting what it can do with a sneeze or a sniff! It's also much less useful if I have a cold, but that's reasonable since my voice changes - I've been tempted to set up a second voice file for when I've got a cold, but fortunately I've not had one since I thought of doing it! Nothing's perfect, but some things can be improvements. I'm a lousy, 4-fingered typist. Dragon's quicker and more accurate, so I can accept the imperfections. Lesley Robertson

Lesley Robertson wrote: > Agreed. I also noticed a dramatic improvement when I upgraded to the > versions above 7. It still can't tell the difference between words the > same (eg knight and night) but it would be unreasonable to expect it to > do so. Proof reading and some correction will always be needed, but I They are supposed to be able to do that (to some extent) by context. ViaVoice was successful. I only had Dragon long enough to set it up and test it for a friend with Parkinson's, but it seemed to work better than ViaVoice. Both are going to have trouble with names of people and places, more so if they are "foreign" names (relative to the language of your software). -- Wes Groleau "A man with an experience is never at the mercy of a man with an argument." -- Ron Allen

"Wes Groleau" <groleau+news@freeshell.org> wrote in message news:T9lrj.3887$0B2.3565@trnddc03... > Lesley Robertson wrote: >> Agreed. I also noticed a dramatic improvement when I upgraded to the >> versions above 7. It still can't tell the difference between words the >> same (eg knight and night) but it would be unreasonable to expect it to >> do so. Proof reading and some correction will always be needed, but I > > They are supposed to be able to do that (to some extent) by context. > ViaVoice was successful. I only had Dragon long enough to set it up and > test it for a friend with Parkinson's, but it seemed to work better than > ViaVoice. > > Both are going to have trouble with names of people and places, more so if > they are "foreign" names (relative to the language of your software). > Only the first time you use them. At work, we use a lot of dog latin for microbial names, at home my Scottish and South African family history provides all sorts on interesting words. Once I've used a word and trained the system, it copes the next time. My office word files even include spatterings of Dutch words (eg the address of my office) without any problems. Lesley Robertson