How can I set DNS (9.5) to capitalise after a new line?
Submitted by gzornenplat on Thu, 05/15/2008 - 15:23.
The problem is that DNS only considers a new paragraph as two new lines, and Word considers it one new line.
It's the most annoying part of DNS for me (except of course that Select That selects the word 'that' and the fact that none of the formatting ever works except in the DNS demo).
I have a Heading1
My Heading
I say new-line or press enter and go to a Normal paragraph. No capitalisation.
How to fix?
And why does saying new-line in a heading give you just a new line and the same paragraph type, and not the next paragraph type as set in the styles?
It doesn't seem all that integrated to me.
Surely *someone* at Nuance must have noticed this when they tested it? Didn't they?
Ian
Oops, do not complain too
Oops, do not complain too fast. NaturallySpeaking is a good program but has of course its flaws.
But the case of New-line and New-paragraph is not something they didn't notice, but something that is deliberately in the program like that.
You can customise the New-line behavior though:
When you go to the vocabulary editor ("Open vocabulary editor"), find (and select) the word "\New-line" (empty written form, spoken form is New-line, you will find this in the top of the middle window).
Then open the Properties window, and select "Use alternate written form".
Then in the choice box select "New line and capitalize next".
Greetings, Quintijn
PS Why would you ever want to give the command "Select that"???
Hi,Thanks for your reply,
Hi,
Thanks for your reply, I'll give that a try. Doesn't that mean that it will change for everything though?
So what you are saying is that 'they' decided that in Word, 'New-Paragraph' will always insert two new paragraphs, and 'New-line' will not be the same as pressing [enter] or saying 'press enter'?
Call me the fool, but shouldn't 'New-paragraph' give you a new paragraph? Just the one?
Why would I want to say 'Select That"? Because, in the help, it says: To correct the last thing you said: Say "Select That" or "Correct That."
(Their punctuation.)
So is that incorrect, then?
Don't get me wrong, it's a good program, but I don't see why complaining that it doesn't work the way they say it does work is 'complaining too fast'. I make my programs work.
Ian
gzornenplat wrote: The
The problem is that DNS only considers a new paragraph as two new lines, and Word considers it one new line.
It's the most annoying part of DNS for me (except of course that Select That selects the word 'that' and the fact that none of the formatting ever works except in the DNS demo).
I have a Heading1
My Heading
I say new-line or press enter and go to a Normal paragraph. No capitalisation.
How to fix?
And why does saying new-line in a heading give you just a new line and the same paragraph type, and not the next paragraph type as set in the styles?
It doesn't seem all that integrated to me.
Surely *someone* at Nuance must have noticed this when they tested it? Didn't they?
Ian
Ian,
Are you a relatively new DNS user? It seems to me that you have a number of fixable issues which mostly involve learning a different procedure or work-around. I think though that for the sake of clarity each issue should be broken out into a separate topic. For example, "formating" covers a lot of ground so we really need specific examples to work on.
I don't have anything to offer on Word problems because I don't use that program, but hopefully someone can take a crack at one or two of the issues you raise.
Good luck,
Bruce
Third-Party Command Option
All of the previous recommendations are good
ones but we would like to add that it helps if you let us know what specific
flavor of DNS you are using because if you are using DNS Pro, you could create
the types of commands that you want. If you're using DNS Preferred, you could
add a third-party command utility such as KnowBrainer
2007 which already includes the commands you asked for. In KnowBrainer we
say Start Line to move the cursor to
the end of the current paragraph (in the event that you are editing the current
paragraph), strike the {Enter} Key one time to move down one line and force the
next word you dictate to begin with a capital letter; even if the previous
paragraph doesn't end in punctuation. You can also substitute Tab Line if you wish to indent your
paragraphs. You'll find additional information in our What is
KnowBrainer PDF file
KnowBrainer wrote: All of
All of the previous recommendations are good
ones but we would like to add that it helps if you let us know what specific
flavor of DNS you are using because if you are using DNS Pro, you could create
the types of commands that you want. If you're using DNS Preferred, you could
add a third-party command utility such as KnowBrainer
2007 which already includes the commands you asked for. In KnowBrainer we
say Start Line to move the cursor to
the end of the current paragraph (in the event that you are editing the current
paragraph), strike the {Enter} Key one time to move down one line and force the
next word you dictate to begin with a capital letter; even if the previous
paragraph doesn't end in punctuation. You can also substitute Tab Line if you wish to indent your
paragraphs. You'll find additional information in our What is
KnowBrainer PDF file
Hi,
Thanks for replying.
I have v9.51.001.009 running on Vista Ultimate with SP1
I've trued DNS before ages ago (version 4, I think) and found it unusable, which is why I went for the Standard version this time.
There are two categories of problems I have with DNS.
Firstly the ones where it doesn't work as it says it will in the documentation - saying 'bold that' prints the words instead of bolding the selected text, for example.
Secondly, it doesn't work as you would expect it would - 'new paragraph' inserting two paragraphs and similar problems.
Work-arounds and kludges are all very well, but I can't help but say that I feel that the program should work as per the documentation. If a kludge fixes a problem in one application but also screws up the functionality in another (like having to change what new-line does so you can't use it elsewhere as per the documentation) it's not much of a fix IMNSHO.
If it were one of *my* applications, I would either fix the program or amend the documentation, but that is just me. It shouldn't matter what version, it should do what it says it does.
On the other hand, it is good to know that I might be able to set up different users for different applications. I already have to save my files on a regular basis because I am getting 'the application is in an unstable state' messages at least twice a day and then I lose all the words I have added in the mean time, so changing users wouldn't be much more of an inconvenience. Before you ask, this also happens on a clean install of Vista on a different machine, so it's not just that the one machine is unlucky.
As for KnowBrainer, I think that I can't justify spending USD200 in order to get a 50-quid program to work.
As for pressing [enter], unfortunately the whole point of buying DNS was to help someone who is unable to press [enter]. And be honest, voice command isn't really voice command if she has to press keys on the keyboard.
Thanks for your help, I shall try as advised and post a couple of other problems.
I *am* impressed with it, and would like to get it to work as it should - or at least, as I think it should - I'm slowly getting the US spellings out of the vocabulary and replacing them with UK ones (this *is* by the way, the UK dictionary
'email' for 'e-mail', 'recognise' for 'recognize' 'dash' being a dash and not two dashes etc)
Thanks again for the responses and suggestions - I appreciate your time and effort in helping me.
Ian
Ian, Based what you've told
Ian,
Based what you've told us, its safe to say your DNS installation is whacko -- one message a year that 'the application is in an unstable state' might be forgettable, but two a day indicates a radical problem.
That means that the behavior of your installation of DNS is all screwed up, so no wonder it doesn't work per the documentation.
The first thing to do is to fix the installation problem before you do anymore work. Its not clear that its a problem with your user (files), or just DNS as a whole, or maybe your PC.
I'm not an authority in this area, so take my suggestion with a grain of salt -- wait to see if KnowBrainer or someone else has a different suggestion.
I would formally uninstall DNS and ditch your old user files. There is a DNS 9.5 cleanser available at Nuance, and it wouldn't hurt to download and use it.
Then re-install the app using all the defaults re: location, etc. Then create a new user and let us know how its going. Its a bit of work, but its worth it to get a usable installation, which you clearly don't have now!
Bruce
Thanks Bruce.Does that mean
Thanks Bruce.
Does that mean I have no choice but to lose months of training the thing to recognise my style (or lack of it)?
It was a clean install onto a new install of Vista with SP1 so I don't know what the difference could be if I do the same again
Ian
Ian, Clean or not, its dirty
Ian,
Clean or not, its dirty now. If you get the same results with a new install, then you know there's a problem somewhere else in your system -- maybe a bad Vista, bad memory chip, etc. Are you having any problems with applications other than DNS?
Installs can and do go bad, although there's no way to know exactly where the problem is. Less doubtful is that your DNS is unusable.
Bruce
Not working properly, for
Not working properly, for sure.
As a programmer, I give all my machines a very hard time. Not-so-good machines need rebooting every couple of hours on Windows, average machines maybe twice a day to be safe. This one is better than average and could probably go for a day or three without rebooting. I don't let it though unless I am working on some other computr and leaving this one just to run background tasks and monitor things.
Eclipse, which is quite good at slowing down has caused no problems. In fact, apart from one program (TextPad) which needed a patch for Vista, everything has been remarkably well-behaved.
I've backed up the user files and am about to un/re-install. I'll let you know. I hate to lose all that effort put into fixing the vocan, though. If Dragon had a switch that said 'Caps On' means 'Caps On' (as opposed to only Capping the words it feels like in titles), that would be good
Oh, and save a record of the words I've deleted, too.
Oh well...
Ian
gzornenplat wrote: As a
As a programmer, Ian
BTW, if you're not already familiar with the VoiceCoder list/open source product, its an essential resource:
http://www.onelist.com/community/VoiceCoder
Bruce
You wouldn't happen to know
You wouldn't happen to know where the cleanser is, would you?
The Nuance search is returning a server error (no, it's not me, it's an http 500 - that's them, definitely)
Ian
ERROR: Unable to fetch the stylesheet athttp://www.nuance.com/css/google_uk.xslt
If you encounter this error in preview, please note that our sample pages are designed only for preview purposes. The links and buttons do not function as in real searches.
If you encounter this error in search, please make sure that the specified stylesheet is either the collection name or a valid URL
My mistake. The current one
My mistake. The current one is only for 9.1:
http://support.nuance.com/downloads/
Bruce
gzornenplat
Secondly, it doesn't work as you would expect it would - 'new paragraph' inserting two paragraphs and similar problems.
This really is not a NaturallySpeaking problem. NaturallySpeaking sets the "new-paragraph" word to insert two returns, because that is what most word processors require. In other words, it is the equivalent of pressing the enter key twice.
The reason you are getting two paragraphs in Microsoft Word is that you are using a style that automatically inserts an extra line between paragraphs.
You have two options to correct this behavior -- you could say "new line" when using Microsoft Word. That's what I do. Or you could change the Microsoft Word style.
Hi Matt,Mmm, I realise
Hi Matt,
Mmm, I realise that, but Dragon say things like 'works with your favourite programs ... virtually any MS program' and 'Get the most from MS Word' and 'works the way *you* work' and they have an MS Word 'context' and a Word section in the help and loads of stuff about Natural Language.
The implication of all this in my mind is that DNS is supposed to work with Word, i.e. it integrates itself with the Word application.
Having to change the behaviour of New-Paragraph to put out a New-Line and cap it is a kludge. And even then it doesn't change the paragraph style according to the rules set up, so a Heading 1 will stay as a Heading 1 even if Word knows to change it to a Normal paragraph style. And it buggers up all the other programs you use.
Look at it from the point of view of a shopper: you go into a store, the box tells you it is designed to work with Word and 'work the way *you* work', so you buy it, get it home, and you can't get a new paragraph without resorting to buying USD200 more software or typing on the keyboard.
I bought it hoping to help someone who is physically unable to type. I'm sure she'll use it with all it's workarounds - she has little alternative, and she's a patient soul - but it doesn't do what it promises on the box, and I know I got her hopes up too high.
Ian
gzornenplat wrote: Having
Having to change the behaviour of New-Paragraph to put out a New-Line and cap it is a kludge.
I did give you instructions how to change New-line behaviour at the start of this thread. The use of New-paragraph doing to times the enter keys is really quite practical.
And even then it doesn't change the paragraph style according to the rules set up, so a Heading 1 will stay as a Heading 1 even if Word knows to change it to a Normal paragraph style. And it buggers up all the other programs you use.
You have a point here, saying New-line when at the end of a Heading 1 should reset to the normal paragraph style.
On the other hand I wouldn't know how it buggers up other programs.
Please be a bit more constructive. Speech recognition is a difficult process, which has been worked upon for several decades. Lots of people can work well with it. Lots of people try to contribute in this community and try to help, also new, users to work with it. But please realise you have to learn a lot before you can use speech recognition as comfortable as you are used to type. By the way how long did it take you to learn to type?
My take, Quintijn
I really do appreciate your
I really do appreciate your point that it is a difficult thing to recognise speech.
It took me six months to learn to type at a reasonable speed, but I had never typed before.
I have been speaking for most of my life, I am a qualified English teacher, and I feel I have got the hang of it now.
This is a speech-recognition program: the onus is on the program to recognise speech, not on me to learn how to speak to it. Not much of that is a problem between me and DNS anyway.
The problem (this problem of new paragraph) is not anything to do with recognising speech. The speech has been recognised correctly ('New-Paragraph').
DNS is aware it is in Word. Dragon know how Word works. They have failed to apply the correct formatting in this context. The sequence should be:
- Recognise 'New-Paragraph' - this is OK (and the most horrendously difficult bit)
- Realise the context is Word - they know this - easy
- Output one CRLF and set caps on for the next word - they fail to do this.
All the hard work has been done successfully, it's the dead-easy it at the end (simply hooking into Word's processing for the enter key) that they fail to do.
*That* is why it is so damn frustrating: All they need to do is output *one* CRLF pair and set caps instead of *two* CRLF pairs and set the caps if we are in Word.
How difficult can that be?
It's like driving 1000 miles to deliver two pizzas to the next-door-neighbour every time the real customer orders one pizza. (Who then has to throw one away or remember to order garlic bread and a coke in order to get a Deep-pan Four Seasons delivered)
It's not a problem of recognising speech, it's doing the right thing once you have.
Ian
gzornenplat wrote:
The problem (this problem of new paragraph) is not anything to do with recognising speech. The speech has been recognised correctly ('New-Paragraph').
DNS is aware it is in Word. Dragon know how Word works. They have failed to apply the correct formatting in this context. The sequence should be:
- Recognise 'New-Paragraph' - this is OK (and the most horrendously difficult bit)
- Realise the context is Word - they know this - easy
- Output one CRLF and set caps on for the next word - they fail to do this.
The discussion on New-line and New-paragraph is longer than this observation of yours. Speech recognition simply treats New-paragraph as two times the enter key. If Nuance would follow your wish it would destroy a lot of other situations people are used to. But you are right, I observed 8 years ago when starting with speech recognition that a paragraph in the NatSpeak way of thinking is something different from a paragraph in Microsoft Word. It should be a nice implementation that if you say "New-paragraph" when - in Word - you are in a Heading style paragraph, it does like you describe. In all other cases (from a Normal paragraph to a Normal paragraph) it should not change its behaviour. You simply have to get used to the NatSpeak way of thinking here.
While the behavior could be corrected, please also be conscious of the fact that New-line and New-paragraph are dictation, not commands. When implementing commands it is easier to taylor to specific applications.
But you completely right: if it is wanted, the NatSpeak programmers should be able to implement this behaviour for Word. But you are wrong stating this is easy. It is a very difficult detail to implement I would think.
Quintijn
You can argue semantics and
You can argue semantics and we'll never come to a conclusion, but when I say 'New-Paragraph' when I'm in Word, I want the program to do (from memory in VB6 it is something like)
newStyle = currentPara.Style.NextStyle
currentPara = worddDoc.Paragraphs.Add
currentPara.Style = newStyle
Sorry to differ, but I used to program Word from VB for folding money in the last century, and it really *is* that easy.
I really don't see how anyone can argue that 'New-Paragraph' should produce two newlines and *never* produce a new paragraph under any circumstances. Fair enough, output two new lines if you are producing a .txt document that has no concept of a paragraph other than a blank line between two others, but if the program is sophisticated enough to have a concept of proper paragraphs, why not give me a new paragraph? Is that really too much to ask?
You seem to be arguing that it is
Ian
Ian, There's marketing and
Ian,
There's marketing and there's reality. Now that your friend has bought the program, she's going to have to deal with reality -- you can quibble (probably means you're a good intervenor!), but she has to learn how to use the bloody thing.
That Star Trek thing is an awful albatross. Writing a program to translate speech into text is tough to do -- were it otherwise, there would be a flock of products. In fact, there are only two SFAIK, and someday Nuance is liable to yank DNS when it can't get enough revenue vs. M$' freebie version.
Its NOT simple or intuitive to use, but most of the problems you've cited have to do with the interface between SR and the target application, which is complicated because there's little standardization among software makers. In other words, the problems you've cited have to do with formating, program interface, etc., rather than accuracy. This type of issue is conditioned by the relationship between the SR maker and the application maker. Nuance doesn't doesn't have the most transparent relationship with M$, but they've managed to get along, however grudgingly.
In other words, the accuracy of DNS (and Vista SR for that matter) is good enough to make it worth the investment needed to master the persistent, persnickety nits.
I hope your friend acquires enough facility with the program that she will feel emboldened enough to join us. In the meantime, we'll try to keep your special role in mind. Conversely, the onus is on you to use your time and energy to find solutions on her behalf rather than bemoan things none of us can alter.
Bruce
gzornenplat wrote:
You can argue semantics and we'll never come to a conclusion, but when I say 'New-Paragraph' when I'm in Word, I want the program to do
...
You seem to be arguing that it is
Ian
I want to argue that maybe most computer users (not VB-programmers) are not aware that when they press enter twice they do not get a new paragraph (in Word sense), but two. Maybe it is Word that introduced a different sense of paragraphs.
Should NatSpeak follow Word or follow intuition of most users?
When you are in normal text, you should do a shift+enter to remain inside a paragraph, but most users feel they remain in the same paragraph when typing enter once (or say New-line).
Technically NatSpeak should in the Word application do a "hard" enter when leaving a paragraph (especially a Heading), so Word can react correspondingly. The "soft" enter apparently does give the expected style change.
For the rest I would like to stop this discussion as we cannot influence this behavior.
Quintijn
I seem to remember this was
I seem to remember this was a common problem in Word 97. I remember solving it but I cannot recall how. However there is some threads on the subject, if not on this site, then in the now-defunct, but still accessible
http://groups.yahoo.com/group/VoiceGroup/,
http://groups.yahoo.com/group/VoiceComp/ or
http://groups.yahoo.com/group/VoiceComp/
Quentin
Capitalize new line, etc.
I've gradually come to the opinion that it's better to separate the functions of voice recognition and word processing. When I'm working from a printed-out pdf document, I almost always dictate in Dragon Pad, which is pretty primitive for a word processor, but just fine as an undemanding interface for DNS. My goal is just to get it all down on paper. (Well, not really on paper, but you know what I mean.) Then I open the rtf file in Word and twiddle with the formatting. I have to proofread the whole thing through carefully anyway, so formatting doesn't really add all that much time to the task.
The exception is when I have a well-formatted Word document (one that uses tabs rather than spaces, etc.) to translate and then I can simply overwrite the source text with English, using the formatting that's already in place. DNS 9.5 is a big improvement over earlier versions for staying on the rails in Word.
WordPad To the Rescue
Another option for quick
down and dirty dictation is WordPad which is nearly identical to DragonPad
except for one wonderful feature. WordPad has multiple undo capability like
Microsoft Word and even Windows while DragonPad will only let you perform
multiple undo of dictation via the scratch that command. WordPad is as easy
to launch as saying Start WordPad. Just food for thought…
Lunis Orcutt - Developer of KnowBrainer &
Host
of the Http://www.KnowBrainer.com Speech Recognition Forum
Absolutely agree with Ian, and yet..
I've been here. I've wished that all voice software would have a couple of modes to choose from.
[x] I know what I'm doing. Please enter 1 new line when I say new line, and one paragraph marker when I say new paragraph. I'll work the Styles in Word to get the effects I want.
[y] I pretty much use Word as a glorified typewriter, and so do all my colleagues. Better put in two carriage returns for "New Paragraph", because that's what we do.
So basically, I'd go for [x]. All the software assumes (grrr!) you're a [y]-type.
However, it's energy-draining to keep running around putting fingers in dykes, so I also got used to a two-pass workflow. One pass, using WordPad, DragonPad or whatever to generate text, and a second iteration, in Word, to tart up the formatting. Which usually meant stripping double CR's so that you end up with one paragraph per paragraph (with or without spacing, but controlled by Styles).
Someone might pipe up and
Someone might pipe up and say: "I wonder how Microsoft will finagle/finesse the interaction between its speech recognition and word processing products?"
The dithering optimist might suggest the speech recognition people would lead the word processing folk to embrace a smarter/simpler interface. A cynic would retort that would be like a hair on the tail wagging the whole dog. That would imply the onus is on the speech recognition folk to do all the work -- they seem to be capable people but I'm not sure they're up to the job.
It could be interesting to see how it works out.
Bruce
PS: Matt, good to hear from your again -- some no doubt innocent British usages have double entendre connotations to Yankee ears/eyes.