Registering and logging in removes this ad.
Registering and logging in removes this ad.
Windows Vista SR compared with XP SR
Submitted by itamarez on Thu, 11/24/2005 - 20:18.
I have posted on my Speech Website and in the FILES section of the Yahoo! MS Speech Group a document comparing MS Speech in Windows Vista with the current Office 2003/Windows XP version. This comparison is based on information gathered from several sources plus my short testing. I hope that the beta version will shortly allow more valid testing of the product.
Comments and corrections are most welcome!
The link to my Speech website is:
http://speech.even-zohar.com/Vista-vs-Xp.htm
or
http://sr.even-zohar.com/Vista-vs-Xp.htm
Itamar Even-Zohar
What I'd like to see is a
What I'd like to see is a full blown comparison between DNS, MS Speech in Vista and ViaVoice.
Who would be the winner and why?
Itamar
Itamar Even-Zohar
ita...@even-zohar.com
Well, you can try and do it yourself. There are some categories suggested in the various documents on my speech Website. You can gather various points from there.
Personally, I have given up ViaVoice, though I liked very much its navigation and dictation macros. I have given up DNS for English (but keep it still for transcription), because it never functioned for me as well as MS Speech under Office 2003 (and I expect it to be even better under Vista), but I have not given up DNS for Spanish and French, because these are the only available applications, and they are very good. I have NOT upgraded to version 8, however, because DNS blocks multilingual use in that version. I therefore expect MS to develop more languages in order to make DNS (ScanSoft turned Nuance, actually) to handle this feature with some more user-friendliness.
Itamar Even-Zohar
As a point of information
As a point of information the blocking of the installation of different versions of DNS seems more in the nature of a ScanSoft "cock-up" rather than a conspiracy.
Chuck Runquist has explained on this forum how it should be possible to get instructions on how to install more than one version of DNS 8.
I would be interested to know if you have any real information on Microsoft's intentions regarding different language versions. So far I have not been able to find out if they even intend providing UK phonetics with the new Vista speech product.
Graham Hendry
Itamar
Itamar Even-Zohar
ita...@even-zohar.com
Graham,
I was not thinking in terms of "conspiracy"; I know it was just indifference and lack of interest. While the original Dragon company devised various language applications to work separately, ScanSoft adopted the VV architecture, which at the same time improved multilingual usage and made it more complicated. However, there was never even a HINT in any of the DNS Manuals HOW one is expected to work with several languages. No one could solve any problems for me at DNS Support (I have quite a bulk of funny correspondence with them), such as the bizarre mixture of languages in the HELP feature.
When I buy an application, I expect it to work immediately. I don;t wish to rely on some obscure information that the manufacturer is not even willing to post on its Website for registered users. I don't want to call a number in Europe (international call) just to find out that the person I am talking to has not got the slighest clue to the problem I raise. Since you live in England yourself (isn't it?), can you try and do what you have suggested, namely call your British support centre and ask them what you can do with your packages? For example, if you have bought the French package, then the Spanish package, what can you do in order to activate English, French, and Spanish (currenly mutually blocking eac other)? I would be happy to hear from you that the support person you called really had the adequate information about the available solution.
BTW, Nuance could solve these problems very easily, if it had made it possible to buy flexible packages, or created a patch to solve the problem (and allowed download purchase of language nodule).
Itamar
Itamar, My comment
Itamar,
My comment concerning "conspiracy" was really just that I generally believe in the "cock-up" rather than "conspiracy" theory of history or of life for that matter.
As far as "blocking" goes in DNS 8 it is not only different language versions that cannot be installed on the same PC - it also applies to different versions such as Medical or Legal. It is the result of the activation mechanism. That said I fully agree that it shouldn't have happened as the problem was pointed out during beta testing.
You are correct that I am in the UK. The Brits, as you know, have a reputation for only conversing with "furriners" in English and very S-L-O-W-L-Y and very loud so it perhaps isn't surprising that information on European language versions of DNS is sketchy here. I have personally only sold one non-English copy of DNS 8 (German) since launch so can't be classed as an expert on multi-language installations. I have, though, shared your frustrations while attempting to order a Japanese version for a client - the conclusion - "don't start from here" - buy it in Japan!
It does seem strange that a company with major facilities in Belgium should be so backward in recognising the need for better multi-language support. There are historical excuses (possibly) in that Dragon Systems had a different license structure for each European country but that was a long time ago. What is much more likely is that re-engineering the software to allow for plug-in language models doesn't satisfy the "bean counters" in terms of return on investment. The number of people who are fluent in a range of languages and who wish to use speech recognition must be very small in terms of total sales. [It would be interesting to know the proportion of sales for each language]
Perhaps in re-engineering DNS to 64 bit Nuance might address the issue of compatibility of different (language) versions.
I have, as a ScanSoft Premier Partner (we don't yet know the Nuance designation), raised compatibility with ScanSoft UK and asked that support people are able to give users accurate information on how to enable more than one version of DNS. I trust that the request has been followed up!
Graham Hendry
www.itspeaking.co.uk
Itamar
Itamar Even-Zohar
ita...@even-zohar.com
Graham,
Thank you for your explanations. It is true that the number of people who use SR for several languages is not high. However, if a DNS package in the Netherlands (and probably Belgium) contains 4 languages, this is at least evidence of some consumer pressure. I cannot imagine a company like ScanSoft/Nuance to simply volunteer such a sale out of sheer good will. If such a package was feasible without complications, why should it be complicated to add one, two, or three more languages? Or why could they not produce a MODULAR package, where the customer can select the combination of languages? I can assure you that the manufacturing of such packages (at least the fixed ones) is not a complicated matter at all. All it takes is some understanding and some willingness. Perhaps one Graham Hendry can tell them that it's time to move on.
You have elegantly evaded my question about how real it is to expect a Support person in Europe to give you a correct answer how to activate more languages. Please tell me that you have indeed called such a person and that the person has understood your question and could provide you with the correct answer. I truly doubt that you might find such a person.
However, even if you managed, I don't think this is the right way to handle this problem. The solution should be available immediately to all registered users. If one buys 2 packages, one wishes those packages to work immediately, not after awkward phone calls which often bring no solution (speaking out of expreince). A simple patch could take care of that.
Itamar
Ah, you are one shrewd
Ah, you are one shrewd moderator/instigator!
Seriously, "winner" might be an unsustainable reification. SR is an extremely complex product with many possible evaluative dimensions. I guess a "winner" would be the one that most people prefer, or the one that scores the most points over multiple evaluative dimensions (possibly weighted) -- which might not produce the same "winner"!
Still, I think a matrix comparing the capabilities of the different products might be useful. Any determination of a "winner" would be inherently subjective and moot -- but I guess that's the point!
Some kind of Wiki-type tool allowing multiple contributions to both the evaluative dimensions and the results for each might dimension be the best approach, but I'm not sure exactly how to do it or which (relatively) unbiased individual might direct the effort. One problem is finding someone who is sufficiently familar with all three products but can resist his/her innate biases -- or at least can acknowledge them to the extent that other participants can correct them in producing a collaborative project.
Bruce
Itamar
Itamar Even-Zohar
ita...@even-zohar.com
Bruce,
Seriously, I have attempted the comparison you were thinking about more than once. In an old document, available from my Speech Website (http://sr.even-zohar.com), I suggested that the comparison be made on the basis of categories. The title of the document is "DISCUSSING THE QUALITY OF VOICE RECOGNITION PROGRAMS IN COMPARATIVE TERMS", and the categories suggested were:
Hardware connectivity, Flexible installation, Training, Learning, Management of memory, Updating and backups, Working with various applications, Transcription, Dictation, Auxiliary features during dictation, Correction, Navigation, Editing, Alternative commands, Macros and vocabulary extension, Languages, Flexible spelling (English), Language varieties -- Pronunciations and accents, Text-to-speech, Documentation.
The applications compared were:
[1] Dragon's NaturallySpeaking "Preferred" version -- U.S. and UK English, Spanish
[2] L&H VoiceXpress Professional version 5 -- U.S. English
[3] Philips' Free Speech 2000 -- U.S. and UK English, Spanish of Northern Latin America, Italian, French, and German.
[4] ViaVoice Millennium PRO (release 7) -- U.S. English.
Most details discussed in that document are no longer valid; moreover, two out of the four applications have disappeared, and one is oscillating between life and death. I have never received any comments to this project, so I assumed it was of no interest to anyone and have had no appetite for updating it.
Next, I made a brief comparative table in a more updated document, a Survey of the currently available SR applications. You can see this attempt at
http://www.tau.ac.il/~itamarez/sr/survey.htm#_Tabl...
Again, no one (not even you) has come with any suggestions for this comparative table. People seem to want simple answers: "Is DNS better than MS?" "Is ViaVoice better than DNS?" You know that these questions cannot be answered unless one is really fanatic about this or that application. The possible answers I belive can only be something like: "depending on what you need and on what you prefer and on what happens to work for you because of your particular kind of speech, perhaps A or B or C would suit you best". I have never recommended any single SR application to anyone, just described my personal experience. Other people have had different experience, as you know.
I would be very happy to try our hands together at making a comparative table of the various available features and describe how well they work. I hope Windows Vista beta will soon be available, so that we would be able to compare our experience with it, too.
Itamar
itamarez wrote:Itamar
Itamar Even-Zohar
ita...@even-zohar.com
.... and one is oscillating between life and death. v
Itamar, you are presumably referring to ViaVoice. You clearly have not been following my comments in this regard. If it were dead, or dying, it would not still be appearing on the shelves.
Susan Fulton has a reasonably up-to-date comparative table on her site, with which I'm sure you're familiar.
However no comparisons have been made using VV 10.5, although comparison with VV 10 would be sufficient.
Personally, I would like to see the Vista version with the best of DNS and VV thrown together, the glitches ironed out, ease of use with transcribers and a functionality which would enable a wide selection of users to use the one user training.
Quentin
Itamar
Itamar Even-Zohar
ita...@even-zohar.com
Quentin,
I wholeheartedly join your wishes as expressed in your last paragraph! I am afraid, though, that won't be achieved without some pressures. However, while no one seemed to be listening at ScanSoft and IBM, at least David Mowatt and now Rob Chambers can be directly approached. This helps a lot, I believe, as I have already seen how flexible their team is and how keenly they listen. But our wishes may be met fully only if MANY users express them.
From what I have seen and read, SR Vista will have some of the macro capacities of VV, the highest recognition accuracy, many of the useful UI features of DNS (you'll admit that DNS' UI is more versatile than VV's), etc., but there will also be some irritating shortcomings like the inability to keep speech data, the absence of transcription, etc. More language are promised, however, and Rob hinted at the possibility that they might implement transcription at a later date.
Itamar
I wonder how Dragon got such
I wonder how Dragon got such an early jump on non-English corpora? I assume that once you've got the basic SR engine programming down, its fairly easy to adapt it to other languages, but getting the corpora for any language is a logistical bugbear.
That raises the question of where/how MS would get its corpora.
Bruce
BruceCyr wrote:I wonder how
I wonder how Dragon got such an early jump on non-English corpora? I assume that once you've got the basic SR engine programming down, its fairly easy to adapt it to other languages, but getting the corpora for any language is a logistical bugbear.
That raises the question of where/how MS would get its corpora.
Bruce
Bruce et al.,
First, Dragon Systems got their break because of their direct involvement With DARPA in 1999. The only two companies to derive benefit from the large corpora of speech data available as a result of DARPA were IBM and Dragon Systems. This corpora was based on more than 170 hours of acoustic data, along with the associated text/language data from Broadcast News (i.e., CNN etc.). This corpora of speech data incorporated the acoustic information and analysis data for over 7 million words (total collected vocabulary & realtime pronunciation not to be confused with unique vocabulary.)Dragon Systems and IBM were included in DARPA 1999 as a result of their defense contracts.
Second, one of the reasons for the merger of ScanSoft and Nuance was the additional large corpora of speech data collected by Nuance.
All the other players in the game have benefited from these early large vocabulary, continuous SR corpora as time has gone on. However, Nuance (ScanSoft) now has the largest and most comprehensive collection of speech data (corpora), and I suspect that they will go to great lengths to protect what they have acquired now that the competition is beginning to ramp up between Nuance and MS.
The corpora of speech data is most critical to accuracy issues. Vocabulary is consistent and based on phonetics. This is the easy part because it is a constant for each language. However, it is the acoustic corpora that makes for accuracy relative to the recognizers (speech engines) and acoustic/language model algorithms (HMM's) in SR.
Chuck
Sorry to have slighted your
Sorry to have slighted your and Susan Fulton's good works by failing to cite them.
I don't have an interest in pursuing this project but acknowledge that it would be useful to the SR community if extant but outdated matrices were updated.
As for which is best, that again is not a subject that I want to attack. The "market" will provide a definitive answer ultimately, but its a peculiar market: Nuance is a direct participant, MS is an "indirect" participant, and IBM is . . . Well, I'll defer to others to interpret crivon1's characterization to their own standards
Your position most closely reflects the reality: The "best" product depends on context of its use. You like MS SR for English, and Nuance products for translation work (there aren't any alternatives!)
I still prefer DNS for English, but I like the fact that MS SR is rapidly approaching what I consider competitive status.
Bruce
Itamar,would you update your
Itamar,would you update your comparative tables.
Under ViaVoice 10, on the first-line it states " Fully supports various applications, but MS Word XP and 2003 are not supported. This should be deleted, perhaps you would also changed the reference to ViaVoice 10 to 10.5. All in the interests of accuracy and avoidance of misleading statements.
Quentin
Itamar
Itamar Even-Zohar
ita...@even-zohar.com
Quentin,
Your claim is a bit of problematic. As far as I know, you may get a patch for making your VV compatible with Office 2003 only if you get in touch with the support department of Nuance (?). Nuance does not offer any downloadable patch open to all, does it? If I am correct, then I don't consider this to be a true support, but I am willing to add the additional information. Please help me formulate a concise line with the necessary instruction to VV owners how to go about it.
Itamar
itamarez wrote:Itamar
Itamar Even-Zohar
ita...@even-zohar.com
Quentin,
Your claim is a bit of problematic. As far as I know, you may get a patch for making your VV compatible with Office 2003 only if you get in touch with the support department of Nuance (?). Nuance does not offer any downloadable patch open to all, does it? If I am correct, then I don't consider this to be a true support, but I am willing to add the additional information. Please help me formulate a concise line with the necessary instruction to VV owners how to go about it.
Itamar
Itamar,
There is no patch as such.
What happened was that either IBM or ScanSoft (now Nuance) developed VV 10 to become 10.5, and, amongst other improvements, the latter included Office 2003 compatibility. A lot of the glitches which had been listed over the years in via-...@yahoo.com were taken on-board and rectified.
I would suggest that you insert the following:-
" To upgrade Via Voice 10 to ViaVoice 10.5:
First of all make sure you have your ViaVoice serial number to hand. The serial number is at the back of the instruction manual. You will be asked for this.
Log-on to the ViaVoice section of Nuance (formerly ScanSoft), go to Support, contact Technical Support, send an e-mail using the Problem Report Form. Tell them that you want ViaVoice 10.5 to replace your VV 10 .
You should then get a response as I did (from the UK Site) as follows:-
"Thank you for contacting Nuance Technical Support regarding ViaVoice Pro. There is now an updated version of ViaVoice available, which supports Office 2003 (ViaVoice 10.5). However, this version is not available for download. If you wish, we can exchange your ViaVoice Version for a new ViaVoice CD free of charge. In order to do this, please send your old ViaVoice 10 CD."
Simply follow the instructions in the response, and it appears that they will exchange your old CD for the new one.
It would appear that US Nuance wants $30 for the exchange in the US. The UK Nuance are not looking for payment (at the moment). "
(Personally I think that the $30 is worth it.)
There was a long thread about this in via-...@yahoo.com, and it can be found this site's Via Voice forum at http://www.speechcomputing.com/node/95.
I have tried it and I can confirm 10.5 is compatible with Office 2003.
Feel free to edit this in any way you want in order to pass the information on.
Quentin
[ this post is a reply to
[ this post is a reply to Bruce's observation that wiki would be a suitable place to collect comparisons between different speech recognition products]
I probably don't qualify as a disinterested party, but here's a starter page: http://speechwiki.org/SR/SpeechComparison.html, all are welcome to add content.
-- Scott W
Speech resources at SpeechWiki
Try QuickSwitch at Applied Recognition today
ScottW wrote:[
[ http://speechwiki.org/SR/SpeechComparison.html, all are welcome to add content.
Any reason why the ViaVoice heading is almost greyed out
Quentin
The ViaVoice heading has a
The ViaVoice heading has a dashed underline which simply indicates that no topic on SpeechWiki exists for ViaVoice. You (or anyone else) can create a topic for ViaVoice by simply clicking on a dashed ViaVoice link and adding some content.
-- Scott W
Speech resources at SpeechWiki
Try QuickSwitch at Applied Recognition today
Itamar, This is very
Itamar,
This is very helpful. Thanks for doing this. Sounds like Windows Vista has some promise, but there are some obvious holes.
itamarez wrote:I have posted
I have posted on my Speech Website and in the FILES section of the Yahoo! MS Speech Group a document comparing MS Speech in Windows Vista with the current Office 2003/Windows XP version. This comparison is based on information gathered from several sources plus my short testing. I hope that the beta version will shortly allow more valid testing of the product.
Comments and corrections are most welcome!
The link to my Speech website is:
http://speech.even-zohar.com/Vista-vs-Xp.htm
or
http://sr.even-zohar.com/Vista-vs-Xp.htm
Itamar Even-Zohar
If anyone is going to provide a competitive/comparative matrix or other analysis of the various speech recognition products currently available, there are a number of considerations that are essential to forming the basis for this type of analysis. I offer the following based on my education (Masters thesis) and background in conducting competitive/comparative analysis of speech recognition products, which I conducted at L&H between 1999 and 2002 as technical solutions and DNS SDK product manager.
1. Any comparative/competitive analysis has to be objective. Performance, accuracy, and related issues have to be based on a standard set of criteria that everyone can agree on. Otherwise, no significant competitive/comparative analysis can be drawn because the results will always be based on the bias of the observer, as well as any bias introduced by the methodology used. This is a difficult area to work with and most analyses in the past have been biased and statistically invalid.
2. Competitive/comparative matrices should be based on “product features.” A comparative/competitive matrix needs to be 100% objective. That is, either the product has the feature (i.e., whether in whole or in part) or it does not. Otherwise, such matrices become mired in subjectivity. In addition, a comparative/competitive matrix needs to provide a quick overview of feature comparisons. If such matrices contain a lot of text and data explanations, they are not very useful.
I would suggest that until or unless they generally accepted standard for statistical analysis of performance, accuracy and related issues are developed, any comparative/competitive analysis in this area should clearly be avoided. As Itamar and others have indicated, whether or not a particular SR product performs better or less well than another is generally based on personal preferences. Anyone who is going to attempt a statistical analysis on these aspects of speech recognition must, of necessity, conform to the basic standards for conducting statistical research. Otherwise, such results would be worse than worthless, they would be mediocre.
I would be glad to provide a copy of the standard matrix that I used for reporting the competitive/comparative features of L&H VoiceXpress, Dragon NaturallySpeaking®, IBM ViaVoice, And Philips FreeSpeech 2000 that I did for L&H. Perhaps this would give everyone a basis for creating such a matrix based on contemporary evaluations of the current SR product market.
Chuck
Itamar
Itamar Even-Zohar
ita...@even-zohar.com
Chuck,
My comparative table for MS had nothing to do with evaluation. I simply tried to indicate what features are available or not available in each of the versions. The picture is not fully clear yet for Vista, and we still can make some strong requests to Rob Chambers about implementing some badly missing features either soon or at a later date.
You know very well that evaluations cannot be either objective or statistically supported. You can only have data about how many people have expressed satisfaction with some product. You can test how many out of a randomly selected group of people achieve high accuracy. You cannot count on their reports, but must have the samples plus results for inspection.
The only helpful comparative table is a factual one, as you indicated. But since the features are not identical, the comparison can mostly be verbal. I tried to make a comparative table with YES and NO (see my 'Survey...") for DNS, MS, VV and the Swedish VX.
Itamar
Itamar, I wasn't slighting
Itamar,
I wasn't slighting your table or attempts. In fact, your approach is what I would site as an example of how it should be done. I was mostly concerned with enlightening the unenlightened as to Comparitive/Competitive analysis 101.
Some of the other sites referenced started to go into performance, etc. I know you understand the process, but a lot of readers to these posts don't.
Chuck
itamarez wrote:Itamar
Itamar Even-Zohar
ita...@even-zohar.com
Chuck,
The only helpful comparative table is a factual one, as you indicated. But since the features are not identical, the comparison can mostly be verbal. I tried to make a comparative table with YES and NO (see my 'Survey...") for DNS, MS, VV and the Swedish VX.
Itamar
Itamar,
Here is the approach that we used at L&H:
1. Take the product with the most features and use it as your left side (comparison) column.
2. List all other products left to right in order of # of comparitive features. Granted, some features are different, but, for example, IBM and DNS have many comparative features that are close enough to find some common ground.
3. Use Yes/No/Partial for the feature support rows.
4. If the 2nd, 3rd, etc. column products have features that the main column (#1) does not have, add them as appropriate. That is, expand the list of features as necessary. It is easier to create a clean matrix in this way, than to try to do it otherwise.
5. Avoid comments directly in the table cells themselves. Add them as superscripts (i.e., 1, 2, 3, etc.) to the Yes/No/Partial entries, and then post these at the bottom under the matrix. You can also emphasize features with comments by making them another color (I use RED).
This way, the matrix is tight and easily viewable. Those who are interested in the explanations can view them at the bottom. However, it is a good idea not to go overboard on explanations as this clutters the matrix. Best to keep the matrix table as clean as possible. In this way it is more readable.
Also, with regard to Windows Vista, this makes it easier for readers to identify what needs to be presented to Microsoft.
If you can keep the matrix clean and one page (i.e., less the explantions at the bottom), then it gives the reader a better overall view of the comparison.
Chuck
Itamar, I noted some of
Itamar,
I noted some of Rob's comments in the Vista section of your table that were very disappointing, and it would appear that they are going to produce an imperfect product in which one will have to download various patches in order to bring Vista speech nearer to the level of DNS and VV. This is clearly a nonsense.
At this stage, there should have been sufficient experience to be able to produce an SR programme equal to, if not better, than DNS or VV, with all the facilities that already exist in them. I believe that they're wasting their time if they're going to bring out an inferior product.
Rob, I trust that your development team are observing this thread and particularly making note of Itamar and Chuck's comments.
There is no question that SR will be the way forward in the immediate future, particularly with the unavailability of secretarial assistance at reasonable cost. This is already seen in large corporations where operatives are expected to do their own typing.
There is no question that with the abundance of facilities available in the principal existing SR programmes, new users can find it daunting. However it is not difficult to learn how to use SR keeping to the basics for a long enough familiarisation period, after which the more complicated facilities can be tackled.
As an example, I have just carried out an exercise I have never done before - to see how many macros I have personally made for use in Microsoft Word, and I found that I have 74. A lot of these were done in the flush of excitement when I said to myself "that would make a very useful macro", and I then proceeded to make the macro, but in many cases have never used them since. I would say that on average I would use and 12-15 of these macros on a regular basis. (These do not include, of course, the built-in Macros in VV). However I would not be without the macro creating facility and this could be a deal breaker as far as I'm concerned, and I believe it would create disappointment to new users who became familiar with Vista and SR.
An all inclusive package is what is necessary.
Quentin
crivon1 wrote: I noted some
I noted some of Rob's comments in the Vista section of your table that were very disappointing, and it would appear that they are going to produce an imperfect product in which one will have to download various patches in order to bring Vista speech nearer to the level of DNS and VV. This is clearly a nonsense.
First of all, Microsoft Speech version 7.1 is FREE. Second of all, just because you use certain features, it does not mean that all people use those features. The fact that they will be there for download or be provided by third parties means they are there if you will need them.
The most interesting thing in this whole realm of Microsoft Speech is that the developers are actually monitoring these user groups and forums. They seem to be listening and are truly interested. I suspect they will also be listening for several years, as they really want their product to work. Can you say the same for IBM (Ooops, I forgot they are not involved in speech recognition for the desktop anymore) and ScanSoft/Nuance? If ScanSoft/Nuance is listening and has customer support it is not apparent.
--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
See us at: http://www.eMicrophones.com/index.asp
Read, "Key Steps to High Speech Recognition Accuracy" at:
http://www.emicrophones.com/docDetails.asp?Documen...
A few things... 1) I find it
A few things...
1) I find it interesting that you think it acceptable to jump through hoops and pay $30 for an incremental upgrade for product that almost certainly will never see a new version, but unacceptable to download add-ons to Microsoft's speech recognition product.
2) Product development is about making choices among finite resources (time, money, people). Choosing perfection amounts to choosing failure.
3) There's a great deal of question as to whether speech recognition will become "the way forward" in the immediate or even distant future.
ScottW wrote: I find it
I find it interesting that you think it acceptable to jump through hoops and pay $30 for an incremental upgrade for product that almost certainly will never see a new version, but unacceptable to download add-ons to Microsoft's speech recognition product.
Effectively VV 10.5 is a new product. There was a two-year gap between 10 and 10.5. The same thing happened when it went from 9 to10.
No one seemed to quibble when DNS went from 7 to 8, and they had to buy a full version at full retail price. Did they not have to pay for the upgrade?
At least with VV, they give credit for having purchased the previous version. Did DNS do that?.
As regards further development, how are the mavens so positive that IBM are not continuing development?
Martin, let us not go down that road again.
As regards Vista speech, why should there not be one package for all?
Quentin
>As regards further
>As regards further development, how are the mavens so positive that IBM are not continuing development?
development may still be continuing on the ViaVoice recognition engine, I just don't think we'll ever see another desktop ViaVoice product release. Let me count the ways:
> As regards Vista speech, why should there not be one package for all?
obviously I don't have any inside knowledge of Microsoft's plans (for SR or anything else) - but I'll guess that this is a means of controlling costs and scheduling. For large integrated software products, the costs of coding any particular feature is just a fraction of its true cost. Officially adding feature means the following:
ScottW wrote: ViaVoice
ViaVoice 10.5 is virtually unheard of and difficult to obtain.
I cannot understand this comment.
ViaVoice 10.5 standard and Pro USB are readily available on the shelves in Ireland at normal retail prices, not discounted.
As far as I'm aware it is the same in the UK.
I cannot speak for the US.
As far as packaging the product been Vista is concerned, the SR technology is already in place with MS Speech, and presumably it is an add-on within Vista , and since Microsoft technology development in this area has been ongoing for in excess of three years what I have suggested surely is not unreasonable.
Quentin