Loquendo supports distributed speech recognition standards

For those of you following DSR developments you might be interested in learning Loquendo's recently announced ASR 6.5 product supports for the ETSI Aurora Advanced Front-End and extensions. They've also included support for the latest W3C SISR (semantic interp. tags) specification.

I haven't found a formal press release on this, nor do I find much on their website, but I did notice a posting in the SpeechTek Magazine's weekly newsblast.

Read the newsblast posting.

IBM's Announces new Speech/Multimodal Tools

Last Friday IBM announced a preview of its Voice Toolkit and Multimodal Tools Plus. The former supports VoiceXML and the recently open sourced Reusable Dialog Components (RDC) and is designed to lower the bar in terms of implementing speech apps. The latter is an Eclipse-based tool that supports implementing X+V apps targeting browser-based apps on embedded devices.

Read the article.

Speech-enabled Media Players

No word yet on exactly which media players will support this technology, but Scansoft's SpeechWorks and Gracenote are teaming up to allow users to interface with their media players via voice commands instead of using their thumbs.

Read the entire article.

MS Speech Server 2004 finally gets press in 2005!

A press release Microsoft issued today on their Speech Server 2004 product is the first press release they've issued on this topic since they launched Speech Server at SpeechTek/AVIOS last spring. Evidently, an "Enterprise Infrastructure Product of the Year Award" presented by Frost & Sullivan has brought them out of the woodwork.

The Microsoft press release comes on the heels of a VoiceGenie press release issued yesterday in which they claim double digit growth in sales in 2004. VoiceGenie of course is a champion of the W3C speech standards (VoiceXML, etc.) while Microsoft's SpeechServer 2004 is for the most part based on proprietary technology.

Wackiest SALT PR of the year!

Hang onto your hats folks - this recent press release from EZ Scores has spkydog putting his bark collar on, and will have you howling as well! Check out some of these quotes:

"EZ Scores had 8 of their best employees attending the show, looking for leads and creating a buzz."

Um, uh, you mean best looking? Shucks, why didn't they just send all nine employees? And just how is it that one creates buzz by simply attending a trade show? Instead of passing out t-shirts and coffee mugs with your company label you hand out cigars of unknown origin? Wait a minute... you need a booth if you're going to hand anything out besides your business card. Hmmm.

"EZ Scores will be promoting its product throughout Jacksonville during the Super Bowl. They will be located at 12 or more key retail locations in Jacksonville Landing and the surrounding area, all within close proximity to the NFL Experience. ...... This is a major achievement for a startup company such as EZ Scores."

Hmmm, at least they qualified the boast! spkydog sure hopes their investors agree.

"The EZ Scores system will automatically update their preselected information just by saying "my teams" or "my players." This is a great feature for any sports fan."

Wait a second, who's talking to who here? The EZ Score system says "my teams" and bingo, your preselected info is updated automatically! Things just can't get ezier... but hang on, they do get wackier!

"EZ Scores has spent the past six months converting its introductory version coded by VXML (voice extensible markup language) application over to S.A.L.T. (speech application language tags). S.A.L.T. has many additional benefits, not to mention that it is faster, cleaner, and more accurate. It allows EZ Scores customers to customize their accounts. "

Ok, folks at this point I hope you have your boots on. So just what feature(s) does SALT have that VoiceXML lacks that lets you write an app that allows customers to customize their accounts? Duh.

SALT is faster? How so? Good grief, the guy on the phone can only listen and bark so fast, right?

Speech reco accuracy has little to absolutely nothing to do with the markup interpreter managing your speech dialog, and much to do with the underlying speech technology the platform happens to deploy.

To say SALT is cleaner than VoiceXML is borderline insane. Take a peek at these code examples from the SALT specification rewritten in VoiceXML if any doubts remain.

This SALT-related press release has some of the wackiest quotes spkydog has encountered to-date. Perhaps second only to the infamous "dead horse" statement Microsoft's X.D. Huang made a couple of years ago. But then again, that's not saying much - spkydog can only remember one or two SALT-related press releases being issued since Huang's embarrassing faux pas. ;-)

Woof! Woof!

ScanSoft's speech reco enhances Nintendo's GameCube

Here's an interesting announcement that came today from ScanSoft. Nintendo has incorporated ScanSoft's speech technology into the Mario Party 6 video game. One notable and necessary feature is that the acoustic models have been trained with children speakers.

If speech reco in video games were to become more common, it will be interesting to see the overall impact in the future in terms of the utility of speech technology in everyday applications. For example, if kids grow up accustomed to talking machine, it seems they will more easily accomodate the limitations in speech recognition technology. For example, watch the typical user in their 60s use a mouse/keyboard to accomplish a task with a computer, and then watch a teenager carry out the same task. Same technology, same user interface, but the teenager will typically be more adept in interacting with the machine, as they've been using the technology in their early formative years, likely before they even learned to read an write!

Read the entire press release.

VoiceXML Forum Certified Platforms

Eureka! spkydog was speaking prophetically in his posting yesterday, when he suggested that eventually some VoiceXML tool vendors would announce their tooling supports VoiceXML platforms that are formally certified by the VoiceXML Forum. This morning a press release was issued by VoiceObjects and VoxPilot doing just that! The press release cites the fact that the VoxPilot platform is independently certified, and that refers to the Forum's platform certification program. VoxPilot was one of the first platform's that obtained the certification and the test results are publically available on the Forum's website.

What's the big deal you ask? Well, if you are interested in portable VoiceXML content, this is a very big deal, as the odds are good that content developed using VoiceObjects Factory, its going to port quite readily to any and all Forum certified platforms - that includes 4 popular platforms to-date, and many more (as rumor has it) that are currently in the certification pipeline.

The bottom line? Developers will get the max flexibility and bang for their buck if they utilize tools that target certified platforms. Tool vendors should insist that their platform partners get certified, and platform vendors shouldn't think twice!

For now, VoiceObjects is the tool vendor to watch!

Read the press release.

More and more VoiceXML dev tools!

Today Envox announced their VoiceXML Studio 6. The claim is that the tool reduces the effort required to implement a VoiceXML app by 50%. According to the specs on the Envox site, the tool works with VoiceXML 2.0 "compliant" gateways and J2EE compliant application servers. Application server compliancy is meaningful, but VoiceXML 2.0 compliancy is still open to interpretation. I suppose at minimum, we can assume the VoiceXML the tool generates runs just fine on the Envox platform.

I'm just wondering how long it will be before these press releases for VoiceXML tools say something like "Supports any VoiceXML Forum Certified Platform"? As the number of certified platform begins to grow, I suppose that will happen, and the first dev tool vendor who can honestly make the claim will be in a pretty good position, as will the developers who use their tools.

Read the press release.

Speech Technology at CES

The 2005 Consumer Electronics Show is taking place this week in Las Vegas. Besides the traditional crashing demos given by Bill Gates, (this year's feature act was a crashing Xbox console) there was plenty of interesting speech technology to be found. Here are a few of the highlights:

Samsung announced two new mobile phones. The a890 apparently builds a grammar from a list of canned short text messages, which the user can utter using speech. The more interesting device is the p207, which is the first mobile device spkydog is aware of that has an honest-to-goodness embedded speech dictation engine. Based on my experience using dictation engines on PCs with lots of memory and 3GHz hyper-threaded procs, my instinct leads me to believe this feature will have about as much utility as the camera on your mobile phone - i.e. a short-lived novelty when you take the phone out of the box. Nevertheless, its a step in the right direction, and has great potential as the technology continues to mature.

Motorola announced a Bluetooth car kit, which accesses speaker-dependent voice tags stored in a mobile handset, to provide consumers with an intuitive in-vehicle communication experience.

Sensory demonstrated its FluentChip (TM), an "ultra-compact" embedded speech solution.

And of course, for a little fun you don't want to miss reading about the new speech reco and other delightful features in the new "Microsoft AI", that Bill Gates announced in his keynote! ;-)

VoiceGenie does it again!

Customer Interactions Solutions magazine just named VoiceGenie's VoiceXML platform the "2004 Producted of the year." I'll agree with that! Not only do they have great products, but they have a great team, and are a model citizen and leader in the VoiceXML community. With brilliant folks like Rob Marchand and others aboard, its no wonder they are going places. (spkydog doesn't work there either!)

Read the full article.

New issue of the VoiceXML Review

A new issue of the VoiceXML Review has been published. Of particular interest is the article on distributed speech-recognition, by David Pearce (Motorola). I expect we will be hearing more about this technology in 2005, now that the industry standards are established, and more ASR vendors are starting to support the technology in their products.