Microsoft Vista to silence IBM ViaVoice, Nuance

SOFTWARE BEHEMOTH Microsoft plans to include speech recognition in Windows Vista. Who will buy products like IBM's Viavoice or Nuance's "Dragon Naturally Speaking" if dictation and voice navigation becomes "an integral part of the OS"?

This is one of several questions I have pondered, after watching a local demo of Vista's speech recognition and dictation which gave me a serious case of deja vu, as IBM was doing this ten years ago with OS/2 Warp 4 and its built-in VoiceType dictation. It took the Microsoft Vole only 10 years to "innovate" this to the market. I have to say that the product worked quite well, and with continuous speech, which certainly shows Microsoft has used the money from the OS proloads and MS Office cash cows quite well in these last ten years.

The main question I had was: is Microsoft "doing a Netscape" on both Nuance's Dragon and IBM's Viavoice products? In other words, how can these companies compete on desktop dictation software if the feature is "integrated" into the OS? To me, it's just like watching "IE vs Netscape" all over again. You can argue all day about the superiority of each of these established voice engines compared to Microsoft's new offering, but in the end, the odds are against them. In the words of the NY Times' David Pogue: "consumers aren't actually choosing Internet Explorer; in fact, they're not choosing. They just use what came on their Windows computers".

Another issue at stake is how this leaves "desktop Linux" at a disadvantage. Just wait until you start to see "Vista vs. Suse", "Vista vs. Linspire" or "Vista vs. Fedora" comparisons. I'm sure there will be an uncomfortable entry in there: "integrated voice recognition": Vista YES, Linux: NO. And oh what irony: one of the companies which pioneered voice dictation and navigation, IBM, seems to have totally forgotten its desktop product: Viavoice, first neglecting it for a long time, then dropping the linux version, and finally handing over the remnants of it - the Windows and Mac OS-X versions - to Scansoft, which later ended up in the hands of arch competitor Nuance, which naturally markets Dragon Naturally Speaking. Why are they supposed to promote IBM's product any harder than the company's own?

Nuance: no Linux support
If you see Microsoft, the 800 pound gorilla coming to get you with the intention of making your main product irrelevant, you might want to follow that old adage that says "the enemy of my enemy is my friend". Hence, I'd have expected Nuance to start looking at non-Windows operating systems more closely - big fish in a small pond vs irrelevantly tiny fish in the big windows pond - but that doesn't seem to be the case, at least for now. Erica Hill from Nuance told the INQUIRER: "At this time we do not offer support for Linux but we are always evaluating new and different options for Dragon Naturally Speaking and listen closely to customer feedback and demand as we develop each new version".

So, desktop Linux users who would like to enjoy voice dictation into e-mail and StarOffice / OpenOffice documents currently have no mature solution to use, thanks to the wonderful wisdom - my words - of Microsoft's competitors. Erica Hill from Nuance assured us she's confident about the company's position, and downplayed the threat of Vista's speech recognition: "While there may be a passing resemblance to the Netscape example that you cite, there is a glaring difference in that we believe Dragon is a far superior product to other speech recognition available today. Having offered our solutions for more than a decade, we've built a loyal base and have found that customers - home users and professionals alike - are only accepting of speech technology that is first and foremost highly accurate and also user-friendly".

Then she proceeded to cite what the company perceives are its product's advantages: "On top of its technical advantages, Dragon is offered in multiple languages, for multiple specialities (i.e. Medical and Legal fields with customized dictionaries), with greater functionality throughout the product line, an all-inclusive box (manuals, quick start guides, headset mic, etc.) and dedicated customer support, which all contribute to a strong, confident market position". I should note, by now, that the demo of Vista's dictation and speech recognition was in Spanish language, and the Microsoft demoperson stated that they're actively "tuning" the engine to each language and locale. Perhaps it's time for Nuance to start worrying?

Nuance's Erica Hill conceded that Microsoft's bundling could cost the company some customers, but continued the verbal assault on Microsoft's product nonetheless: "there will clearly be some percentage of users that find the Vista speech recognition acceptable. However, it's likely a greater percentage will try it, like it and look for something more powerful. I think this is very similar to imaging technology as well. Microsoft includes a scaled-back document imaging utility in Office but our Imaging business is still very profitable because with scanning and OCR - as with speech - people demand accuracy, quality and features that simply are not included in a free version from Microsoft. Our $75M and growing imaging business is proof of this! In fact, we are perhaps more concerned about the forthcoming launch eroding consumer confidence in speech recognition given recent public displays of Vista. Now the onus is upon us to underscore just how good Nuance speech technology is and how much people can benefit from Dragon".

Meanwhile, IBM sleeps...
David Brigida at IBM confirmed that IBM's current Windows and MacOS products are now in the hands of its main competitor, Nuance, after it merged with Scansoft -now that's what I call wisdom, or lack thereof-. He confirmed however that the "Viavoice for Linux" product is not part of the distribution agreement with Nuance-Scansoft, so IBM could - again, my words - decide to distribute it either as freeware or to open source if, if they wanted. David said: "You are correct, Nuance distributes and markets the Windows and Mac OS version of Viavoice only. With regards to Viavoice for Linux as a matter of policy IBM does comment on future product plans. So there in little I can share with you about any plans for that product".

The irony of this is that IBM's OS/2 Warp 4.0 had built-in voice navigation AND dictation bundled in the OS ten years ago and pioneered the concept. And remember that OS/2 Warp had to deal with hardware that's 10 years older and much slower. That takes some talented programming.

Am I delusional by thinking that Nuance could make even more money by having its flagship windows product ported to linux and bundled with every copy of, say, Novell Linux 11 or the next version of Linspire, just to mention two flagship desktop linux distros? Or am I crazy in thinking that if IBM released Viavoice for Linux under a freeware - or better, dual GNU/Apache license allowing both pure opensource and also mixing with proprietary software - it would help Big Blue sell an awful lot more copies of Linux on the desktop, by matching the features of Vista in this critical area?

I think that Microsoft currently has the upper hand on voice integration, and that Nuance and IBM are in trouble, not to mention the lack of this handy feature in desktop Linux which puts the platform at a disadvantage. Novell, IBM, Red Hat, are you guys listening? µ

See Also
Nuance Scansoft thinks Microsoft voice technology poor
Open Sourced IBM speech recognition code doesn't include ViaVoice
Dragon Naturally Speaking gets close(r) to perfection
CMU Sphinx Group Open Source Speech Recognition Engines