Registering and logging in removes this ad.
Registering and logging in removes this ad.
The Relationship Between an App's Market Share and SR Support of It
Submitted by BruceCyr on Fri, 08/26/2005 - 02:12.
I'm starting a new thread because the old one was so frayed as to be unreadable.
Here:
http://www.speechcomputing.com/comment/reply/302
lem729 said:
I agree IE is the easiest with Dragon, with its "click link" setup, and then "choose 15. That's because Dragon (for whatever reasons)--I could be cynical and think money is involved 
-- was designed to operate with IE6.
"Cynical" isn't the word I would choose. If you think MS would pay DNS to build in support for IE, I would suggest you don't understand the market.
At the time DNS8 was being planned and finalized, IE had well over 90% of the browser market. If DNS didn't support IE, that would be a strong reason for a prospective SR user NOT to buy it.
There probably are some "arrangements" between MS and Nuance (nee ScanSoft), but I'm not privy to them. In particular, I suspect MS might have needed access to some of Nuance/ScanSoft/Dragon's patents etc., while N/S/D might have needed access to some of MS' OS and application secrets, but that's just speculation.
In any case, DNS' developers had strong market incentives to make NatSpeak as compatible with MS' apps as possible -- its simply a matter of the numbers of market shares -- MS's got them up the wazoo, and the others don't.
FWIW, its hard work to build in command and control and Select and Say SR support for an application -- you have to know a lot about an application's programming conventions and interfaces. Since programs change regularly, and since MS is the reigning maestro of programmed change in their programs as a means to keep up the revenue stream, the SR apps have to change too. That means regular programming costs.
Combine that with the fact that SR is a sliver of a niche market, and you can see that there is little incentive for the SR manufacturer to overtly support any program that doesn't have a large market share -- say at least 25%. Mozilla and its variants and Opera together still fail that test by a wide margin.
Bruce
Is there something
Is there something precluding Dragon (Scan Soft) from working with the developers from more than one browser. I mean Microsoft may have a large majority of the market share, but there still are millions of users of the other browsers.
I'm not saying Dragon doesn't have to be damn sure it works with IE, which has the biggest market share. That's a given. It does. It's really whether it can do more to make the other browsers work better, without needing to expend inordinate energy in doing it, because it just isn't beyond its means anymore. Dragon knows the technology. I mean, can't it help develop Natural Speak macros also?
It's of course speculation (with a history, though, to make one naturally suspicious), but isn't it possible that Microsoft elicited some commitment from Dragon that it would limit how much it supports the other browsers (for a financial arrangement), so that the leading voice activation system did a significantly better job with IE? That would enhance IE's user share, and damage the competition. Perhaps, this wasn't even explicit in the relationship, but implicit . . .
Now I'm suggesting that, only because that kind of behavior by Microsoft would not be inconsistent with past Microsoft actions in trying to dominate the browser market, and create a monopoly.
Remember how Microsoft dealt with Netscape. The courts thought Microsoft's practices were illegal/monopolistic, but by then Netscape was near dead.
Just look at something basic like the MSNBC.COM home page.
http://www.msnbc.msn.com/
I DO NOT THINK it would be hard for that page to be designed so that Firefox and Opera could display video from it? It might even make Microsoft look better if they did that. I mean, surely Microsoft knows how to design a web page. Some web designers might be ignorant or lazy (not have the cash or want to do the work to make their page work with the alternative browsers. But Microsoft (which could easily make the video on these pages read in other browsers) and has reams of cash to be able to do it doesn't seem to want to.
"You said: "it's hard work to build in command and control and Select and Say SR support for an application -- you have to know a lot about an application's programming conventions and interfaces." And you would know more about that than I. Maybe, Dragon can't do more in terms of other browsers. I'm just not convinced. If I'm dubious about Mircosoft's motives, it's only because of the way Microsoft has acted in the past.
By the bye, I don't know the acronym "SR user."
"SR" = "speech
"SR" = "speech recognition"
Nothing "precludes" NaturallySpeaking from working with other browsers; rather, it's a waste of its developer's resources to do so. It's more than a matter of writing macros. In particular, enabling Select and Say involves intimate knowledge of an application's programming interface and custom programming to make it work. Firefox is an especially difficult case for NaturallySpeaking because it is written in Java rather than the Microsoft interfaces.
I can tell you don't have much exposure programming because you don't realize how much it costs to do it well -- you seem to think it's merely a matter of waiting a magical technological wand. In fact, it's gritty intellectual work that requires a certain doggedness along with cleverness. In addition, you need first-rate management to keep the programming oriented towards its goals. In short, it's a significant business expenditure.
The constant upgrade game, which is the major source of on-going revenue streams for software businesses, generates a dynamic that constantly changes the competitive terrain. It isn't a "do it once and be done with it" thing, its a constant uphill slog under rainy conditions. That's why its a significant expenditure for any SR maker to maintain both dictation and command and control compatibility with major applications, let alone the host of minor ones which often eschew common programming standards for their own quirky approaches.
Sun, in particular, a major strategic competitor of Microsoft, eschews virtually all of MS' programming interfaces for its own Java approach. That's why its so hard to maintain compatibility with Java applications, like Firefox.
It doesn't make any sense for Microsoft to pay Nuance/ScanSoft money to include support for IE. Until just recently, Microsoft had well over 90% of the browser market, which means that expending further money to expand its penetration would have been a waste of resources. In other words, there would be no business reason to pay for exclusive support of its browser amongst SR users -- a point that is driven space even deeper by the following paragraph.
I don't think you fully appreciate this fact: SR is a sliver of a niche market, that is, it's not worth anybody's effort to pay for market penetration for a mainstream product like a browser. I would guess that the total number of SR users worldwide is less than one million, and that includes all products and all language versions. Contrast that with the fact that there are probably more than 150 million browser users in the US alone.
The main reason why people are interested in desktop SR these days is because it is a showplace for demonstrating competence in the technology for the market where the action is concentrated, namely, commercial speech response systems.
In sum, the argument has nothing to do with Microsoft's history, character, or intentions. It has everything to do with common sense and strategic thinking under the current conditions. In short, this is a clear case where Occam's Razor applies: Why manufacture convoluted and virtually unverifiable conspiracy theories when common sense and strategic analysis suffice? Microsoft is bad enough without us losing our grip on reality and conjuring up paranoid delusions. If we keep our wits we are far more likely to develop an effective counterstrategy.
Bruce
PS: I relegate to a footnote your discussion of displaying video on the MSNBC.com site, because it's an irrelevant red herring.
You do make good points,
You do make good points, Bruce, though I wouldn't relegate to a footnote my discussion of Microsoft's unwillingness to design its web page for other browsers to be able to use as an irrelevant red herring." It's not (from my perspective) that insignificant.
It does evidence--no?--Microsoft's calculated intention to try to dominate the browser market,and fear of giving any competitor a foothold. If as you noted, until just recently Microsoft had well over 90% of the browser market," and had no interest in expending money to expand its penetration," why their interest in blocking the video function on their web page (which they want to be a major internet portal), to alternative browsers.
Rather than suggesting confidence, this unwillingness to design a web page that other browsers can access suggests Microsoft's fundamental obsession with maintaining their domination, even during this period of time when they seemingly had total domination.
Microsoft's past practice in trying to accomplish such domination (which courts have described as, I believe, monopolistic and illegal) give signficant basis for the concern about Dragon's relationship with IE, to the exclusion of working with other browsers. Now, I hear the practical concerns you've raised, one of which is that Firefox works on a Java platform and that would make it much harder for Dragon to work with Firefox (because somehow, you seemed to suggest, IE was not Java based. I may have misunderstood there. In that regard, though, you noted: "Sun, in particular, a major strategic competitor of Microsoft, eschews virtually all of MS' programming interfaces for its own Java approach. That's why its so hard to maintain compatibility with Java applications, like Firefox."
And really, I'm not an expert in how much this java issue precludes Dragon from being able to work with the Firefox or Opera browser, so as to be able to enhance VR for it. But my understanding is that Microsoft reached a litigation compromise (a bit more than a year ago) with Sun, and that they were going to be working/together, collaborating with the vitural Java Platform. http://www.microsoft.com/presspass/press/2004/apr0...
And isn't IE using Sun Java today? I mean I have a Sun java software that operates in my version of IE.
As to your statement that voice activation is only "a sliver of a niche market," maybe . . . today . . . And even so, that feels understated to me. And for tomorrow, very definitely, I would think, "no." Here's a quote from a 2003 web page by speechtechnology.com: "While it isn't completely mainstream, it (VR)will be a part of your lives whether you like it or not. I am sure you have already experienced some of the first generation speech recognition telephony systems. Speech Recognition and Voice Recognition WILL BE EVERYWHERE in a matter of time (emphasis added).
"http://www.speechtechnology.com/techreview.cfm
Speech recognition is getting so "awesomely" good (I'm dazzled with how nice Dragon works for me with Opera and Firefox, Bruce . . .) (it wasn't even fundamentally designed with those browsers in mind) that people WITHOUT DISABILITIES could easily find Voice Recognition Software useful/valuable, particularly in a business context.
I mean why kill yourself typing if you can talk into a computer, which makes only 5 percent errors???? today (in the future, maybe even far less than that). For business, isn't it great for a person to be able to dictate into a microphone, and have that transfered into notes in the computer through voice recognition. We're not talking about a disability need, but a "growing" present and forthcoming business need." I think there's a huge potential market here.
In the past Microsoft has talked about wanting to have voice activation as part of its basic operating system. You know to have it in something like Longhorn, or a followup system. Maybe, they've scaled back their plans, but . . . Indeed, Microsoft is fascinated by VR. Their Voice Command (pocket PC) voice activation system won a major annual computer award.
http://microsoft.handango.com/PlatformProductDetai...
If VR were "a mere sliver of a niche market," would Microsoft be putting so much thought and energy into it? I mean, they see down the road . . .
As to your view that since Microsoft dominates the browser market (and of course, technical issues involved with working with the other browser developers), it only makes sense for Dragon to customize its product for Microsoft, why, there have been something like 60 million downloads of Firefox in the last year os so... That's a very very significant market. The view is that there are a lot less Opera users (than Firefox) (but those figures are deflated some, because Opera can present in 5 different modes (only one of which makes it appear as Opera when the browser goes to a web site. (As an aside, often the web site can identify Opera as IE or FF, because some of Opera's five operating modes can make it appear as IE or FF. (That's because web sites are designed for certain browsers (and may only operate if they think the browser is, for example, IE), but not another browser. Other site will recognize IE and FF, but will not render properly for Opera, unless Opera comes in, in an IE or FF mode. The process by which web sites can identify the browser on the site is called browser sniffing.
http://reviews.cnet.com/4520-6029_7-5818202-1.html...
So some of the statistics for Opera users may be understated (because web sites identify an Opera browser, as an IE or FF browser) (Opera also can appear in a FF mode). Indeed, in Europe, there's a much higher percentage of Opera users than elsewhere. In some countries there, I've seen figures where as high as 30 percent (of all browser users) are believed to use Opera. Scan Soft is an international company . . . No?
Now you suggest Dragon has limitations in being able to get its voice recognition engine to work better with Firefox: "I can tell you don't have much exposure programming because you don't realize how much it costs to do it well -- you seem to think it's merely a matter of waiting a magical technological wand."
I'm no doubt naive there. I've been wrong on many things before. Why, if you called me Mr. Wrong, you might be right.
And I agree, I do not have programming experience. Still, what about these third party developers who are giving us "Hit a Hint" and Mouseless Browsing in Firefox." Is it costing them millions upon millions to accomplish that? I doubt it! Because where are they making "big" money out of it. Bruce, it may not be as difficult as you suggest for Dragon to do more for Firefox, Opera and Netscape, and Dragon might actually be able to do a better job of it than these third party developers.
Maybe as you say, it doesn't make any sense for Microsoft to pay Nuance/ScanSoft money to include support for IE. Still, there can be
"under the table" understandings in business relationships. The reason Microsoft is speeding up its delivery of IE 7 to the public is, in all likelihood, because it's concerned about its collapsing market share. (Isn't that a given?) Do you think it's not aware of the millions of Firefox downloads. What's so paranoid or strange (did you not suggest I was paranoid regarding Mircrosoft, as to suggest that Microsoft might want to continue the relationship it's had with Scan Soft/Dragon in the past (where Dragon Naturally Speaking is designed so that VR works much better with IE), and that it has some under the table understanding with Scan Soft to accomplish that.
I'm just saying that there's a past history with Microsoft, and I for one, will not simply assume that Dragon's exclusive relationship with Mircrosoft is all "above board," and free of business pressures that one company can put on another. Microsoft has leverage because it still dominates the browser market. It can use that leverage to discourage Scan Soft from working with the competitors. Money doesn't have to change hands. A company that once attempted to create a browser monopoly might easily try again. Do (technological leopards change their spots?
lem729 wrote: You do make
You do make good points, Bruce
If only that were always true! Too often my points make as much sense as the meeting in "Hustle & Flow" between DJay and Skinny
[No, wait! That's how my business meetings go! Never mind.]
Basically your post:
1. Tacitly accepts my arguments regarding the business reasons why DNS has no incentives to support non-mainstream software, although it makes diversionary asides to deflect the gravamen of those arguments.
2. Devotes its main energy attempting to rebut my claim that SR represents "a sliver of niche market". Those of us who have used and debated SR for a long time believe that that claim is precisely the reason that delimits the expansion of SR beyond what appear to be its current, natural limits.
3. Finally, returns to the "MS as source of doom and death" theme, like a Wagnerian leitmortiv.
I accept 1 with gratitude, ignore 3 because its a worldview as immune to argument as an article of faith, and focus on 2 because it is substantively most germane.
As to your statement that voice activation is only "a sliver of a niche market," maybe . . . today . . . And even so, that feels understated to me. And for tomorrow, very definitely, I would think, "no." Here's a quote from a 2003 web page by speechtechnology.com: "While it isn't completely mainstream, it (VR)will be a part of your lives whether you like it or not. I am sure you have already experienced some of the first generation speech recognition telephony systems. Speech Recognition and Voice Recognition WILL BE EVERYWHERE in a matter of time (emphasis added).
"http://www.speechtechnology.com/techreview.cfm
This quote concerns commercial use of SR in voice response systems, which, as my note noted, is the hottest and most profitable area for SR. But the focus here is on general recognition desktop SR, so this quote is not pertinent.
Speech recognition is getting so "awesomely" good (I'm dazzled with how nice Dragon works for me with Opera and Firefox, Bruce . . .) (it wasn't even fundamentally designed with those browsers in mind) that people WITHOUT DISABILITIES could easily find Voice Recognition Software useful/valuable, particularly in a business context.
Offer your services to the Nuance/ScanSoft marketing people, who seem to have hit a brickwall after gargantuan promotional efforts -- or even IBM, for that matter, who gave up on the desktop market to focus on the commercial SR market
Seriously, I will address this point below.
I mean why kill yourself typing if you can talk into a computer, which makes only 5 percent errors????
You have lot of room for improvement -- like other long-term, expert SR users, my accuracy NEVER goes below 98%, and is about 99% when I'm in good form!
In the past Microsoft has talked about wanting to have voice activation as part of its basic operating system. You know to have it in something like Longhorn, or a followup system. Maybe, they've scaled back their plans, but . . . Indeed, Microsoft is fascinated by VR. Their Voice Command (pocket PC) voice activation system won a major annual computer award.
http://microsoft.handango.com/PlatformProductDetai...
If VR were "a mere sliver of a niche market," would Microsoft be putting so much thought and energy into it? I mean, they see down the road . . .
As I noted, MS and other SR makers use desktop SR as a showcase to springboard into the commercial market where the real prospects for profit lie. You may note that MS doesn't attempt to make money off its desktop product. Could it be that that market is so small as to be negligible to a shrewd, exploitive business like MS?
I see you also are confused by the difference between Voice Recognition/Activaton and Speech Recognition, although at times the two phrases are used interchangeably by less articulate speakers. Basically, the former phrase refers to a biometric capability, i.e., recognizing someone by the unique characteristics of his/her voice (e.g., gender for one!); while the latter refers to the semantic interpretation of speech to produce, amongst other things, written text. Garage doors, dogs and opera singers, for example, all have recognizable voices, but only people speak symbolicly with semantic intent.
I'm no doubt naive there. I've been wrong on many things before. Why, if you called me Mr. Wrong, you might be right.
And I agree, I do not have programming experience.
Well I can't dispute that! Sorry -- when you open the barn door that wide you have to expect an animal to go through it
Still, what about these third party developers who are giving us "Hit a Hint" and Mouseless Browsing in Firefox." Is it costing them millions upon millions to accomplish that? I doubt it! Because where are they making "big" money out of it. Bruce, it may not be as difficult as you suggest for Dragon to do more for Firefox, Opera and Netscape, and Dragon might actually be able to do a better job of it than these third party developers.
Well, that's true of the open standards/free software communities -- SFAIK all Firefox/Thunderbird extensions, like the products themselves, are free for the download. Most of the extensions appear to be the products of hobbyists and/or people dedicated to improving humankind by providing useful software for free -- no kidding. Trying to sell software is considered tacky if not down-right unethical in the free software community. Of course, not all the players have the same expectations. For example, the big players (i.e., promoters of grand concepts like "open standards") like Sun and Mozilla ultimately aim for some sort of return on investments -- they wouldn't be viable businesses if they didn't have strategic plans. But many of the small players appear to be true-believers of the free software ideology. They even have a patron, if secular, saint, viz., Richard Stallworth.
But open standards make the difference -- anyone who wants can "nominate" him/herself to make an extension. Mozilla makes all the programming code and interfaces free. Its just the opposite with proprietary software, like Opera, for which there aren't many (any?) add-ons, free or for purchase. Moreover, I cite yet another cost I omitted that you don't seem to be aware of: ongoing support. If you add code to speech enable an app, its certain that a proportion of users with ecah app won't be able to use the capability as intended -- especially if its proprietary software with private interfaces, which covers the majority of software.
You can argue this point all you want, but it reflects ignorance of fact and practice rather than a substantive argument -- if DNS stood to benefit from such additional efforts, they would have done them. Economics has the concept of the efficient agent, which is relatively valid when speaking of a profit making entity like the modern corporation. [Not so, as research has revealed, of individuals, who in economics as in all other phases of life don't seem to manifest the "efficient rationality" posited by economic theory.]
As an agent like Nuance surveys its prospects, it compares them to determine which ones are within its capabilties and mazimize returns relative to inputs. In the past Nuance offered greater compatibility with other applications, but in recent versions it has reduced that compatibility. That suggests to me that they have determined from experience that they make more net profit with this reduced profile than with the expanded profile. Maybe its partly due to arrangements with MS, but that argument is undermined by the fact that they have tried to maintain their compatibility with WordPerfect -- there must be enough WordPerfect users to justify this expense.
Lets put the onus on the other side: Why won't Mozilla (Mozilla, Thunderbirs) or Sun (Open/StarOffice) enable their software products to have compatibility with NatSpeak? Well, partly because that would mean using MS' proprietary but public interfaces, which would mean MS could negatively impact these applications by arbitrarily changing these standards. So why don't Mozilla and Sun offer to make the adjustments to enable NatSpeak compatibility? What's precluding them from doing that? Why, because they can't make any money that way. Does that make them any less "evil" than MS? No, it just makes them less fortunate and less resourceful about scrounging for new business opportunities.
You see, the game you play can be played from the opposite side with opposite conclusions! It all depends on where you start, i.e., on where your loyalties and biases take you. You can argue MS is an agent of death and doom, but it might also be that the structure of the capitalist system dictates the way things fall out, and I suggest that's what's true here.
Look: I'm not bothered when people say they hate MS -- I'm not that fond of them either. It may be that they are going to move into a competitive place in desktop SR, or maybe not once they've demonstrated their bona fides and entered the commercial market. They may also to try to make a pitch for developing turnkey SR systems for specific organizations or industries -- something the early developers of discrete recognition systems also tried, and at which they failed because their technology wasn't good enough.
However, we have to recognize the SR market if finite, and that apparently we can see its boundaries. There are two reasons:
1. Not that many people need or want to either dictate text or use verbal command and control, which naturally limits the reach of SR. The proportion of people who produce text as part of their work is very small: writers (of whom journalists appear to be the largest portion); programmers; professionals like doctors, psychologists, lawyers, policeman, etc., and ancillary professionals; academics, translators, transcriptionists; etc. Surely I'm omitting some key but SMALL group of SR users. The point is that the raw numbers just don't add up to much of a cut out of the general populace of PC users.
But only enlightened and/or disabled minorities of these delimited sub-populations want/need to use SR. A very few able-bodied professionals use SR as a productivity enhancer. I suspect the larger majority of users comes from those of us who are not able-bodied. There have been heated discussions in the community about the reasons why more able-bodied potential users don't adopt the technology, and while there is no consensus on the cause, the fact is clear: They don't and probably won't unless they are galvanized out of their inertia. I know several SR-using, producivity-enhancing attorneys hereabouts who have tried to proselytize their colleagues without success. [They may have failed for want resourcefulness. For example, hiring Yakuza-like associates, with their modus operandi, might quickly lead to adoption of SR for producing text!]
2. SR development is in hiatus. Current SR is the product of a techonology that has reached its limits. It isn't going to get much better no matter how much faster or more talented PCs become. SR researchers have said that what we have is what we're going to get for a long time to come because there are no new SR technologies on the horizon. Yet any progress beyond current levels will probably require software new technology. Merely making current alorithms run faster won't do the trick because IBM, for one, has tried that using its supercomputers.
Its as unrealistic to expect a major SR breakthrough in the next decade as it is to attribute the sum of all ills and evils in this field to MS. I think the way to deal with the current technology as well as the current business environment is to assess it realistically and deal with as rationally as possible.
Bruce
You are totally wrong on
You are totally wrong on this! (Hah-ha, that was just to get the header of my post to be provocative, as I know how this forum picks up on the first few words of one's post).
No really, Bruce, as usual, you are very erudite, and it is a pleasure to hear/learn from you. Of course, I'm not going to change my anti-Microsoft bias. I think it's grounded in sound history, but no point beating a dead--heh-heh--horse here on that one . . .
Now you seemed to focus in your latest post on the idea that speech recognition technology was a sliver of a niche market with very limited room for improvement in the future. You pointed out how you get 98 to 99 percent accuracy in your use of Dragon, (disparaging, perhaps, the meagre accuracy I got), but then specifically suggested that its (speech recognition's) future development was quite limited through the next ten years. In that regard, you said:
"SR development is in hiatus. Current SR is the product of a techonology that has reached its limits. It isn't going to get much better no matter how much faster or more talented PCs become. SR researchers have said that what we have is what we're going to get for a long time to come because there are no new SR technologies on the horizon. Yet any progress beyond current levels will probably require software new technology. Merely making current alorithms run faster won't do the trick because IBM, for one, has tried that using its supercomputers. Its as unrealistic to expect a major SR breakthrough in the next decade as it is to attribute the sum of all ills and evils in this field to MS."
Now I'm puzzled by why you think the market for speech recogntion is so limited. Why, with 98 to 99 percent accuracy you get, businesses ought to be chafing at the bit to improve the productivity of their employees? And we're talking about millions of employees worldwide, who could benefit from such accuracy figures. I mean most people talk far faster than they can type.
The issue is not so much that the business world doesn't want to spend on these systems, but rather that in the noisy business environment (high amplitude of background noise), they don't get anywhere near the "awesome" accuracy rate you get, and because of that, are less inclined currently to spend on voice/speech speech recogniton technology. I mean, in the business environment, becaue of the lower recognition rates gotten there, people can still produce more by keyboard and mouse.
One point to consider is that it may not be necessary to improve on the 98-99 percent accuracy rate you get for wider business acceptance, just to be better able to block out background noise--that is for the Voice recognition technology to be able to better distinguish between the speaker/user and that background noise. A very recent Wikipedia article commented on this problem in the noisy business environment, noting a:
"lower recognition accuracy. The typical achievable recognition rate as of 2005 for large-vocabulary speaker-independent systems is about 80%-90% for a clear environment, but can be as low as 50% for scenarios like cellular phone with background noise."
http://speech-recognition.ask.dyndns.dk/
And so why is what you referred to as this "sliver of a niche market" locked in as such? The issue is to find a way to get people in business to be able to use their voice applications in a more noise free setting (or to create some kind of blocking software . . .) that can distinguish between the speaker (voice recognition? and the background noise.
As to your pessimistic--doom and gloom
--view of the future in terms of improvements, neither IBM, nor Bill Gates seem to agree. Both have invested heavily in improving speech recognition, short term. And see (maybe they're naive like me), the following glowing future for Speech Recogniton. In that regard, "IBM claims to have put one hundred speech researchers on the problem of taking ASR beyond the level of human speech recognition by 2010. Bill Gates is also making very large investments in speech recognition research at Microsoft. At SpeechTEK, Gates predicted that by 2011 the quality of ASR will catch up to human speech recognition." http://speech-recognition.ask.dyndns.dk/
Wow, impressive claims!
Now let me digress a bit, and go back to an earlier post of yours in this thread, where you seemed to focus on how financially infeasible it was for Scan Soft to work with other browsers (emphasising Firefox), because its reliance on Sun Java. Your exact quote was: "Sun, in particular, a major strategic competitor of Microsoft, eschews virtually all of MS' programming interfaces for its own Java approach. That's why its so hard to maintain compatibility with Java applications, like Firefox."
You didn't, in your latest post, address my point about Microsoft collaborating as part of its court settlement with Sun Microsystems on Java, What I was saying is that Microsoft now uses Sun Java, and is working/collaborating with Sun on Java (or so the court settlement I alluded to in my earlier post, ostensibly decreed . . .). There may be a simple answer there (though I was hoping you would respond, as I admit there are areas where I can learn a lot from you . . . (as I was perhaps missing a forest for some [java] trees . . .)
In any event, I'm just wondering if it isn't too glib to just dismiss the feasibility of Scan Soft making money with Firefox (and its millions of users (60 million downoads the last year). And query? Is it that crazy for a business developer to work with a browser with a large market, and then one with a smaller (although not insignificant market)? I mean, doesn't IBM's Via Voice have a Windows product and one for MacIntosh. A MacIntosh based browser is probably far far far more unlike a Windows based browser, than Firefox (with its Sun Java) is unlike IE (with Sun Java???? also). And certainly, the number of MacIntosh users is far far far less substantial than the number of Windows IE users? And so, why assume that Scan Soft couldn't make money (by selling its product to Firefox afficonados)? even as IBM Via has attempted to do with Windows IE and MacIntosh.
Now as you can no doubt glean from what I've previously said, I would just like Dragon to not be, in supporting one player (browser) to the exclusion of others, furthering a monopoly. I think (and I've argued this before in this forum) it's in all of OUR interests (who are not employees or owners of Microsoft) that even after IE 7 comes out, the browser field is viewed as still competitive--with a number of good products out there. Competition is what drives business to seek to improve their products. Monopolies can cause them to go to sleep! And Microsoft's IE6 did for years, until they started to see their market share start to dissipate a bit . . . .
In sum, I'm arguing that, not only is it in our interest that the browser field remains competitive. However, it's important that we not let passion be father to our thought. Therefore, I offer up the following cogent--eh?--ideas:
(1) The future for voice/speech recognition may not be as limited in terms of improvement as you suggest. The market (whatever it may be today), is one (as Bill Gates and IBM seem to think) filled with potential, not a mere "sliver of a niche market" for the indefinite future.
(2) IBM, through its, Via Voice seemed to think it was ok to develop a product for a Windows environment with a huge market share; and for a more niche environment like MacIntosh. If they saw potentional profit there, why couldn't Scan Soft see potential profit in the millions of alternative browser users,
e.g., Firefox, Opera, Netscape;
(3) You suggest how technically difficult it would be for Scan Soft to work with and understand another browser (such as Firefox) because of the Java issue. Whether that issue is or is not (to borrow a phrase of yours regarding something I argued for) "a red herring"
if Via Voice could work with IBM Windows, and MacIntosh (two totally different operating systems, and thought there was an economic incentive to do so, why couldn't Scan Soft do the same in terms of, let's say, working with IE and [an] alternative browser[s] (unless somehow there was something about it's relationship with Microsoft, under-the-table, that precluded it from doing so.
As to what I say here, I realize there may be some unintended mischaractierization/error, however modest (which of us purports to know all?) but urge the following of Chaucer's admonition: "be wise and take the grain but leave the chaff."
lem729 wrote: The issue is
The issue is not so much that the business world doesn't want to spend on these systems, but rather that in the noisy business environment (high amplitude of background noise), they don't get anywhere near the "awesome" accuracy rate you get, and because of that, are less inclined currently to spend on voice/speech speech recogniton technology. I mean, in the business environment, becaue of the lower recognition rates gotten there, people can still produce more by keyboard and mouse.
Being the former Technical and SDK Product Manager for DNS, I can emphatically state that the issue of noise is not as much related to accuracy as it is to user attitudes. Back before DNS 6.0 we were able to get 95% accuracy in environments as high as 93+ dB (Comdex) and used to demonstrate this ability regularly at trade shows. In fact, you can, with the proper hardware, get close to 98% accuracy in such environments with the current version of DNS.
The issue is that users and business are "uncomfortable" using SR in the work place for a number of reasons which they put off on "noise." This is a usability issue, not a noise vs. accuracy issue. People are just plain uncomfortable using speech recognition in other than quiet environments. L&H usability studies brought up about 15 reasons why users would not use speech recognition in the workplace. Accuracy was not one of them, and accuracy is not an issue even in some of the noisiest environments. I used to dictate regularly in a cubicle environment at L&H and never had a problem with either accuracy or with productivity. I could always dictate more accurately faster than I could type, and corrections were much easier anyway.
(3) You suggest how technically difficult it would be for Scan Soft to work with and understand another browser (such as Firefox) because of the Java issue. Whether that issue is or is not (to borrow a phrase of yours regarding something I argued for) "a red herring" if Via Voice could work with IBM Windows, and MacIntosh (two totally different operating systems, and thought there was an economic incentive to do so, why couldn't Scan Soft do the same in terms of, let's say, working with IE and [an] alternative browser[s] (unless somehow there was something about it's relationship with Microsoft, under-the-table, that precluded it from doing so.
Unfortunately, your point is an apples and oranges argument. The reason that Java is difficult to work with in SR applications is that SAPI, at least the current versions, do not directly support it. Yes, you can dictate in a Java environment, but Select-and-Say is not SAPI compatible.
Back when we were developing the revised SDK for DNS there were two developmental aspects on the table. The first was to incorporate the L&H VoiceXpress Grammar Studio and related utilites into DNS. This would have given developers a simple straight forward method of creating Natural Language Commands (i.e., not for the average user) and would have allowed multiple commands and command aliasing (i.e., Add an outside border and an inside border, and make the outside border double could be reduced by training to "Add special borders" or anything the user wanted it to be.). This was a feature of VoiceXpress command grammars. The second was to develop a set of special Java DNS extensions that would work with SAPI 4.0a (i.e., create a Java grammar extension for SAPI). Both were scrapped because there was not time in the development lifecycle to incorporate these features and because at the end of the life of L&H, their SAPI Guru and primary SAPI/SDK engineer left to start his own company. The bottom line is that working with Java-based applications requires additional utilities and SR extensions. These are neither easy to develop, and are definitely not market priorities. As I write this I'm currently holding in my hand the technical specifications which I rode for the DNS SDK Re: Java. I can tell you that what would seem like a logical and fairly straight forward integration of Java-based applications by the average user, is really an extremely complex R&D problem.
As to the supposed limitations that Bruce points to, your assessment is basically correct. SR has not reached the limitations of the technology. The statistical algorithms used in SR are being improved each year. Granted that part of the problem is the distinction between computers and human speech capability. That is, people are better at speech recognition than computers for so many reasons it would be a treatise in itself to describe them here. Nevertheless, the problem is not with the models used as much as it is with the compiling and analyzing of speech data. Much of the current speech data comes from the DARPA research of 1999. Very little has been acquired and incorporated between then and now. In short, given the nature of Hidden Markov Models, or HMMs, as well as the other statistical methods used, the problem is one of GIGO. The amount and sophistication of the speech data available has more to do with improvements than the degree of sophistication, or lack thereof, of the models and methods themselves. The extent and quality of the corpus of speech data determines how accurate the models are at transcribing speech to text. This is one reason why ScanSoft acquired Nuance (i.e., Nuance has a large corpus of up to date speech data).
Nevertheless, some care has to be excercised in terms of making assumptions as to where SR is going. Microsoft has been very smart in biding its time and making small advances. They have understood the nature of the average user of speech recognition much better than anyone else has. Their strategy is not driven by marketing hype, it is driven by usability analyses. They are gradually improving their speech recognition software step-by-step being concerned not with the market but for the impact on the user. Having been in on the ground floor of the SR revolution and been involved in many upper-level management discussions on these issues, particularly where Microsoft is concerned, it is quite clear to me that they are waiting for the appropriate time to bring to the surface of full-blown SR application competitive with DNS at all levels. They are not interested in competing with ScanSoft or IBM at this point, and that is not their goal. Nevertheless, speech recognition will continue to improve, and, in fact, the next 10 years will see dramatic improvements in all areas of speech recognition, particularly large vocabulary continuous speech recognition.
Good stuff to know. I stand
Good stuff to know. I stand corrected on both the technical and strategic issues from someone who is much closer to both than I am.
Bruce
Chuck, Thank you for your
Chuck,
Thank you for your very knowledgeable "take" on this. Still, a few issues we've been discussing in this thread puzzle me. I'm definitely not technically proficient in Java, or even computers, am simply trying to understand from a layman's perpective why Scan Soft might not work on its SR software with more than one browser. Yes, the market may be better with IE at this precise moment, but there are millions of people who prefer the other browsers, and they seem to be a market of sorts, if not an opportunity for good public relations.
Now Bruce had originally suggested in this thread (an earlier post) that a reason Scan Soft can't do more with alternative browsers is that they are Sun java based, and Microsoft is not. His exact quote was: "Sun, in particular, a major strategic competitor of Microsoft, eschews virtually all of MS' programming interfaces for its own Java approach. That's why its so hard to maintain compatibility with Java applications, like Firefox." You said, seemingly in support of that:
"The reason that Java is difficult to work with in SR applications is that SAPI, at least the current versions, do not directly support it. Yes, you can dictate in a Java environment, but Select-and-Say is not SAPI compatible . . . The bottom line is that working with Java-based applications requires additional utilities and SR extensions. These are neither easy to develop, and are definitely not market priorities."
I take it from the above--and I apologize in advance if I'm totally discombobulating your thought
--that you are saying that "Select-and-say" for purposes of the "Speech Application Programming Interface" (SAPI) may not be compatible with a a Sun Java environment. And since Scan Soft would want to accomplish "Select-and-say" in its Dragon software, the SUN Java environment is difficult to work with.
But query: haven't the dynamics changed a bit since you were "Technical and SDK Product Manager for DNS?" (I'm not sure if you're affiliated with Scan Soft/Dragon any more, and if not, for how long you haven't been . . . ). I mean what I referenced in an earlier post in this thread was that Microsoft reached a litigation compromise (a bit more than a year ago) with Sun, pursuant to which they were committed with Sun to working/together, collaborating with the latter's vitural Java Platform.
http://www.microsoft.com/presspass/press/2004/apr0...
Certainly IE is making use of Sun Java today? For example, I have on my computer at this instant, Sun Java's platform 1.5, version 1.5.0_04. Under the menu bar Tools in my Internet Explorer 6.0 is the Sun Java Console (where this version is listed as operative for my IE. I believe this version of Java also is operative on my Firefox and Opera browsers.
Thus, I have been arguing in this thread for a few thoughts/ideas, one of which your earlier post seemed to confirm for me--to wit, that the future of SR is not near an impasse, but rather quite bright. My other issues, though are:
(1) It seems that given the current environment where, as part of its litigation settlement with Sun, Microsoft has agreed to collaborate with the latter in using Sun Java, and where I can use it on my IE browser, Scan Soft, if it wants to develop voice recognition software with IE, would need to be more up "up to speed" today, than pre-settlement, on how to make "Select-and-Say" useable in this Sun Java environment. In your earlier post you may have been thinking about when you worked in Dragon as Techinical and SDK Product Manager. But perhaps, the context from when you worked, thus, with Dragon has changed since the Microsoft/SUN settlement litigation.
(2) And if Scan Soft will have to be more "up-to-speed" with Sun Java in dealing with Microsoft's IE product and speach recognition software, then dealing with Sun Java as utilized in some other browsers might not be any more as vexing an issue/problem, particulary browsers that have millions of users (Firefox: 60 million downoads this year), and which, thus, might be considered a potential good market;
(3) Even if if a company may (to develop a certain product) need to work in a somewhat difficult environment--having to learn new things--sometimes, it's worthwhile to do so. Seemingly in that regard, you suggest that Microsoft's:
"strategy is not driven by marketing hype, it is driven by usability analyses. They are gradually improving their speech recognition software step-by-step being concerned not with the market but for the impact on the user."
Isn't it possible that Scan Soft in looking at the millions of people using alternative browers could reach a similar view (that maybe short term there's not money in it for me, but . . . ? I mean AOL gave away their initital software, and people thought they were crazy, but ultimately (before more recent dire turns), they developed an empire. There's something to be said for good public relations, which Microsoft can garner by (as you said showing,in its developement of speech recognition software) less concern about the "market" than about "impact on the user)," and which Scan Soft could very well enhance for itself ("good public relations," that is) if it didn't enter into an exclusive relationship with Microsoft at the expense of the alternative browsers (in other words, if it didn't act in a way, the effect of which could be furtherance of Microsoft's movement back towards a "browser" monopoly), and
(4) An example of a company doing something difficult (even as, let's say Dragon's being able to accomplish "Select and Say" with Speech Recognition software in a Java environment) might have been for it difficult when you worked tih Dragon is IBM with its Via Voice. It offered a Windows product, and another SR product for a much much smaller MacIntosh market (and one where, since the Operating Systems are quite different, presented hurdles possibly???? even far greater than Dragon may've had in getting "Select and Say to function in a Java environment. Possibly, IBM decided that short-term the profits might not be there, but longer term it was taking a gamble (because ultimately who knows where the computer market will go). And maybe, for it, there were public relations considerations also.
In sum, I would still argue that it's in Scan Soft's interest to NOT develop a speech recognition product that gives IE a leg up on its browser competition, that there's enough millions of users interested in the alternatives to IE, that a possible short term loss (for Scan Soft) could be more than offset by good public relations, and the possibility that the expertise Scan Soft can develop working with other browsers could ultimately lead to substantial profits. I mean can Scan Soft or any of us be certain which is the browser of the future? Sometimes, it can help to have your finger in several pots. And finally, given IE's collaboration with Sun on its Sun Java platform, maybe Scan Soft has to learn more of Sun Java anyway, today (just to work with Internet Explorer 7), so working with the other browsers may not be as difficult a hurdle as when you worked as Technical and SDK Product Manager for DNS with Dragon.
I hope the above doesn't seem too muddled
In terms of your last post, Bruce wrote: "Good stuff to know. I stand corrected on both the technical and strategic issues from someone who is much closer to both than I am." The thing is that Bruce is more technically proficient than I am. I read your post, and no question that you are far closer to these issues than I am. Still, as a layman, I can only react to what I understand. I have presented questions, that, quite possibly someone with more technical expertise might not have had.
For me, it's still hard to follow why Scan Soft can't do more with the alternative browser market, both as a matter of pure economics, and good public relations. I mean, can this Java issue be a killer if IE, itself, is now moving towards greater use of Sun Java?
PLEASE DON'T READ THIS IF
PLEASE DON'T READ THIS IF YOU HAVE SOMETHING IMPORTANT TO DO.
I wonder what your background is? I gather that you are retired. I ask because I wonder what you understand about strategic or day-to-day decision making.
For example, could you define "good public relations"? And posit means both to measure and profit from it in a purely business sense? What is a unit of "good public relationships" and how much is it worth to a business?
For example, the AOL case you cite suggests "good public relationships" are as evanescent as intestinal gas and as apt to turn sour when exposed. What value is "good public relationships" if its so volatile?
Bruce
115 words
DONT RESPOND UNLESS . . .
DONT RESPOND UNLESS . . . you can deal with the heart of my prior post (because I was trying to understand something in it). I'm hoping this thread can be educational for me and others. The question raised there was: "doesn’t Microsoft’s enhanced involvement with Sun Java as a result of its court settlement mean that Dragon, in working with IE7, will have to become more familiar with making its software functional in a Sun Java environment? If so, then it may be able to provide, without enormous expenditure of energy, greater functionality than it currently does with alternative Sun Java browsers, such as Firefox.
Now as an aside, AOL had its moment in the sun, and if it frittered it away more recently, I think your conclusion, although verbally brilliant
"good public relationships" are as evanescent as intestinal gas and as apt to turn sour when exposed"),is overstated.
Bill Gates of Microsoft gives billions to charity.
Pfizer gave nearly $450 million in money and medicines last year to charity.
Merck gave $341 million.
“A public opinion poll, conducted by Harris Interactive, found that nearly eight out of ten respondents said they take corporate citizenship into account when deciding whether to buy a particular company's product. About seven out of ten said that corporate citizenship is a factor in their investment decisions. A separate study, commissioned by the The Council on Foundations, found that corporate giving also increases the admiration employees have for the companies they work for, which, in turn, builds employee loyalty and reduces turnover.”
http://www.pfizer.com.ph/corporate/news/generous.h...
There can be a corporate benefit (financially) to being a good citizen. I thought Chuck’s point interesting, that Microsoft is “gradually improving their speech recognition software step-by-step being concerned not with the market but for the impact on the user.” Maybe down the road, they see profits, but they’re not focusing short-term exclusively on market.
Given Microsoft's prior history of monopolistic practices with its browser, an exclusive relation between Scan Soft and Microsoft can very much hurt Scan Soft from a public relations perspective. I, therefore, think Scan Soft, if it can feasibly, should focus on offering a product that benefits fairly equally a broader universe than just the Microsoft browser.
DON'T READ THIS IF YOU HAVE
DON'T READ THIS IF YOU HAVE SOMETHING IMPORTANT TO DO
Over the past seven decades surveys have uniformly determined that approximately eight out of ten US citizens say they support the US Constitution, but when asked if they support specific parts of it (n.b., in the original or in paraphrased terms), majorities decline to support specific provisions. In other words, survey data is irrelevant to the issue at hand, i.e., operationalizing the concept of "good public relationships".
Only by defining "good public relationships" in specific terms can you determine whether a particular action, e.g., generating Select and Say and command and control support for a particular, resistant browser used by no more than a comparative handful of SR users, would be worthwhile to Nuance.
It isn't the case that DNS doesn't NOT support your Opera or Firefox, but rather only that it does not support Select and Say and full command and control in those apps. The reasons it does not support them are:
1. Those applications are written with interfaces that are resistant to the interface that NaturallySpeaking uses to communicate with apps, i.e., apps used by the vast majority of PC users.
2. Not enough SR users use those applications to make it worth the effort to Nuance
If I'm one of the 90+% of DNS users who browses in IE, then I probably don't give a fig whether or not DNS supports other browsers. However, if I were a touch clever, I might actually suspect that the attention given to those other browsers decreases the attention given to supporting my browser since I recognize that Nuance's programming resources are a scarce commodity apportioned among competing demands.
If I'm one of the relative handful of non-IE using SR users, I will be happy if NatuallySpeaking enables Select and Say capability in my non-IE browser. However, since I already have a copy of NatuallySpeaking and have to use it any case because ViaVoice also doesn't support my browser, and of course MS Speech will certainly never support my browser either, I won't be buying an extra copy to share my love with Nuance. In fact, I'm probably going to upgrade to the next version I can afford whether it supports my browser or not because I'm more likely than not a fully or partially handicapped user who needs SR to function.
Your argument presumes that Nuance is a slackly run organization with gobs of undercommited resources it can devote willy-nilly to any one of a number of activities that please its workers. On the contrary, most business always have less of most of their resources (e.g., people, equipment, technical expertise, space, time, etc.) than they have identifiable needs. This dissonance forces them to prioritize their needs according to the basic profit-maximization principle and apportion their resources accordingly. They use hierarchical control to eliminate centrifugal individual impulses that don't adhere to the centralized resource allocation plan, i.e., they fire day-dreaming slackers.
Some people proffer lists of goals and activities they deem suitable for the business from their perspective, i.e., to maximize their own, subjective utility. This type of thinking is of course a prejudice based on ignorance of how businesses work. Its almost as hubristic as a business person who tries to impose profit maximizing principles on, say, education, government or religion.
I'm letting you off the hook of defining "good public relationships" because I've just proven that the action you propose can't possibly generate enough of any return to justify the action no matter how you define the term. I've made this argument before in different terms, but I maybe numbers literally don't count for you. Yet its all a numbers game -- not the grandiose generalities about millions of this and millions of that that you waft about in vague yet majestic sweeps.
What count are relative numbers, conditional probabilities, strategic thinking, contingencies -- basically the very texture of business and life. You may blow these factors off to your heart's content, but eventually people might wonder if your head is in some sort of vision-blurring haze.
Bruce
PS: I note you used a trick header: There weren't any substantive issues in your post! Nevermind, I have enough weight to carry the argument for both of us!
711 words
"DONT RESPOND UNLESS . . .",
"DONT RESPOND UNLESS . . .", I said in my prior post, "you can deal with [which your latest post hasn't] the heart of my prior post: "doesn’t Microsoft’s enhanced involvement with Sun Java as a result of its court settlement mean that Dragon [Nuance], in working with IE7, will have to become more familiar with making its software functional in a Sun Java environment? If so, then it may be able to provide, without enormous expenditure of energy, greater functionality than it currently does with alternative Sun Java browsers, such as Firefox." I was not saying you couldn't address that argument, because the answer wasn't there. I wasn't suggesting that. Just asking for you to address it for my benefit, and for those other users of the forums here, who don't purport to be technological heavyweights, and would like to learn something.
I guess in making any final decision, we'd have to sit on Nuance's board, assess the cost of developing software that had greater functional benefit for browsers other than IE, against the benefits that could accrue to it (in increased sales of Dragon, increased technological know-how, whether there might be the potential for better public relations, whether Nuance should care . . .), etc. I'm not privy to its balance sheet.
IBM found it worth its while to introduce Via Voice for, not just with a Windows based application, but the far smaller MacIntosh. But then Nuance is not IBM. Microsoft is working on voice activation, with a strategy, "being concerned not with the market but for the impact on the user." It's thinking longer term. But then Nuance is not Microsoft. I understand that.
Your argument, Bruce is a market argument, but even there, if someone has 10-15 percent of the browser market, that can still be millions of users
disparage it, however much you choose to do so.
I've been meaning to throw
I've been meaning to throw in my two cents for some time, but have been swamped lately. I also gotten a little bit lost in the discussion, so the below isn't a specific reply to any point or person but just some general comments on the technology side.
1) Mozilla Firefox isn't written in Java. The lack of speech recognition support in Firefox isn't due to the incompatibilities between SAPI and JSAPI, but rather because there is no explicit code supporting it and the Mozilla developers have chosen a development path that won't let them get speech support (Select-and-Say) for free.
Generally speaking, if you wish to provide Dragon NaturallySpeaking Select-and-Say support for your application you have two options, you can use the Dragon API and explicitly code support for it, or for the edit areas you can use Microsoft's standard EDIT or Rich edit controls. (more info at CustomEditControls)
I imagine the Mozilla team doesn't use EDIT or Rich Edit controls because they built their own for cross-platform compatibility. But there's no reason that a developer, either from ScanSoft, Mozilla, or third-party couldn't code in explicit support - it's simply a matter of time and money.
2) A way out - application support for speech recognition input is similar to the situation with printers back in the day. Remember when you were shopping for a printer, and you had to look to see if your word processor supported it - it was a horrible mess. It was bad for software manufacturers, it was bad for printer manufacturers, and was bad for users. Today users buy printers based on the primary qualities of a printer, print quality, cost, pages per minute, cents per page, etc. The idea that your software won't be able to print to any particular printer is absurd. That's because there are well documented interfaces that printer manufacturers can use to ensure that their printer is compatible with Windows (and MacOS), and there are well documented interfaces that software application writers can use to ensure that their software will display on screen and on a printer. Manufacturers and software developers aren't bogged down by the cost of developing testing and supporting N different printers/software packages. Everybody wins.
Microsoft is attempting to the same thing for text input, the Text Services Framework is designed to create a common interface for text input devices, whether they are keyboards, software-based virtual keyboards, pen tablets, speech recognition, gesture pads and devices, eye scanning hardware, or the neural plug. I have no way of knowing what ScanSoft or IBM's plans are for their speech recognition products regarding the Text Services framework, but I sincerely hope they support it.
It would also be great if the Mozilla group implemented TSF interfaces for Firefox, which would give Firefox full dictation support for Microsoft's speech recognition.
-- Scott W
Speech resources at SpeechWiki
Signup for the QuickSwitch Beta
Thanks for clarifying the
Thanks for clarifying the underlying programming issues -- a little technical expertise goes a long way.
I have heard you discuss Text Services before, and I'm gradually coming to understand what that means and its ramifications. The printer driver analogy strikes a bell -- I remember when my most adventurous programming efforts were in developing printer drivers for my word processors and printers. At least some WP makers, like XyWrite, afforded you this opportunity. For others, you had to hope someone had discovered a way to hack it, like inserting hexadecimal codes into the application itself. And of course you had to find a printer maker who documented its codes.
But the key point you make, and the one I've also been making, is that it takes time and money to bring about this compatibility -- its doesn't coalesce out of the air or a cloud of "general public relationships".
Bruce
PS: The QuickSwitch Beta seems to work very well on my machine. Would recommend others try it if they would like to have the conveniences it affords, especially Preferred users who can't role their own code.
I guess from somewhere back
I guess from somewhere back earlier in the thread, I was following the argument pesented that there was a "java" problem in terms of why Nuance couldn't so something more in this area for Firefox, responding to it. You indicated, Steve, that "there's no reason that a developer, either from ScanSoft, Mozilla, or third-party couldn't code in explicit support - it's simply a matter of time and money." At least, it's possible. Anyway, thanks for helping out--finding the time to chime in--on this issue. Microsoft's Text Services Framework sounds like something that could be really helpful here.