V (as in VXML) for Standards Victory

Paul Korzeniowski publishes an interesting article entitled "Speech Recognition Software Slowly Making Progress" in TechNewsWorld. Although published just this morning, Paul has one detail incorrect in his assessment of the VoiceXML vs. SALT issue. While one of the original members of the now defunct SALT Forum, readers should note that Intel is not exactly "solidly in Microsoft's corner" with regard to speech standards. Their recent upgrade to sponsor member and their seat on the board of directors of the VoiceXML Forum must have somehow escaped Mr. Koreniowski's attention.

Voxeo Achieves VoiceXML 2.0 Platform Certification

This morning Voxeo officially announced that their VoiceCenter VoiceXML platform has mustered formal certification by the VoiceXML Forum. Combining the legendary Motorola VoxGateway, one of the first and best performing VoiceXML interpreters on the market, along with the brilliance, energy and voice platform know how of the Voxeo technical team, makes VoiceCenter one of the best VoiceXML deployment platforms available on the planet! But don't take spydog's word for it - go try Voxeo's Evolution 2005 developer portal yourself at http://evolution.voxeo.com!

Read the Voxeo press release.

JSAPI 2.0 under public review

A recent posting inPaul Lamere's "Duke Listens" blog summarizes the highpoints of the JSAPI 2.0 specification (JSR 113) currently under public review. Among the more interesting is the replacement of JSGF and JSML with W3C SRGS and W3C SSML. The specification targets CLDC 1.0 and MIDP 1.0 and above but runs on J2SE as well.

"Opera browser beats the stuffing out of Firefox"

Opera's CEO Jon S. von Tetzchner seems delighted with the fact that million's of Internet users have switched to the Firebox web browser. According to von Tezchner, Firefox's rapid snap up of 5% of the browser market is proof in the pudding that the dominance of Microsoft's archaic Internet Explorer ought not be taken for granted.

For folks interested in speech technology, the recent release of Opera 8 is of particular interest, in that it is the first commercial speech-enabled multimodal web browser on the market. As we have blogged earlier, Opera 8 implements the XHTML+Voice specification. While you might not fully appreciate the utility of equipping your web browser with a speech modality on your desktop, spkydog predicts once you experience it on your UI-constrained mobile device, you'll be barking to a different tune!

Voice-Icon Application: Talk yourself into a job!

"Imagine recording a commercial for prospective employers to listen to in your own voice, where you talk about your strengths and achievements, setting yourself apart with poise and confidence. Voice-Icon is a subscription service that features the ability to record and share up to three unique Voice-Icons with prospective employers, who will be able to listen to your Voice-Icon using a toll free number."

The application is written in VoiceXML and hosted by Invores Systems.

Read the entire press release.

Opera's April Fools Joke

On April 1 Opera issued a formal press release announcing a "P2P Revolution", complete with a demo. P2P human speech that is. Rather entertaining. Check it out if you haven't already.

Read the press release.

Speech Tech and the Grid

Here's an interesting read written by Sumit Badal a Nortel engineer. Badal suggests some possible changes to the architecture of speech algorithms in the context of grid computing to support unconstrained speech recognition (i.e. dictation)

Read the article.

Non-acoustic sensors detect speech without sound

DARPA is working on recognizing speech by using non-acoustic sensors that detect speech via the speaker's nerve and muscle activity, rather than sound itself. While the current applications are thought to be military, eventually civilians might be able to avoid irritating others when using their mobile phones in public. The conversation can be mimed and the "silent" speech recognized and synthesized by the system.

Read the New Scientists article.

Speech Technology Plays Major Role in New Search Engine

From an InternetNews.com article published yesterday:

"Search technology provider Blinkx is madly inking television content deals, signaling a strategy to become the preferred front end for major media companies.

Blinkx's Chandratillake said his company takes a different approach to video search by not only indexing meta information about video content but also parsing the content, so that users can go directly to a relevant portion of the video.

Context Clustering Technology indexes content on the fly using automatic transcription technology, which transliterates content straight from the cable box or from video already stored on the Web. It then brings phonetic matching speech recognition technology to bear on it, identifying the main ideas in segments. He said this method provides better results than relying on closed captioning transcripts, as Google and Yahoo do."

Read the entire article.