W3C Labels VoiceXML 2.0 A Recommendation

rjray writes with news that yesterday "the W3C gave their official stamp of approval-- Technical Recommendation status-- to VoiceXML 2.0. Trade publications covering this include internetnews.com. The move also recognizes the Speech Recognition Grammar Specification (SRGS 1.0), and combines the two together under the heading of 'Speech Interface Framework.'" Here's the W3C press release.

Oh God...

[BlueCodeWarrior]

...is used by developers to describe end-users responses to spoken prompts.

I can only imagine the ramifications for the "adult entertainment" industry.

Re:Oh God...

[Jonboy X]

Ah, I assume you refer to the emerging XXX-XML standard.

Let the games begin!

[ferralis]

...in about 3 or 4 years.

Still, like VRML, it's a beginning. At least now we have some standards to build on. :)

Looking forward to wrestling this beasty to the ground and looking for applications in my own work.

This helps out accessibility...

[tcopeland]

...there's a good discussion of VoiceXML and various other accessibility enablers by Charles McCathieNevile in this presentation [w3.org].

It talks quite a bit about RDF and the Semantic Web [semwebcentral.org], too.

Some confusion in the article

[Anonymous Coward]

While the article is talking about VXML & SRGS becoming official standards, it also talks about how speech recognition "technologies are robust enough to distinguish a person's individual accent or variation". Well, of course that has nothing whatsoever to do with the standards mentioned. It has to do with what the underlying ASR software can do.

Similarly, just because someone wrote a spec for HTML doesn't mean that a browser capable of rendering it existed.

pretty obscure standard

[jilles]

IMHO this is yet another w3c standard that almost nobody uses or will use. I looked through the list of eight known voicexml implementors. The only familiar name I see there is Motorola. The rest of the bunch is small telcos and voice product companies. No doubt there's a niche for this kind of thing and its nice that the w3c offers a platform for industies and organizations in this niche to negotiate some industry standards.

However, my main objection to the word 'recommended standard' is that the w3c publishes these things like crazy without anyone bothering to implement them. A recommended standard from the w3c has little or no authority because you are unlikely to find 100% compliant products (or even 80% compliant).

Something is not a standard because the w3c recommends it but because key software products implement them in an interoperable fashion. For many w3c recommended specifications this is simply not true or at best only partially true. This is especially true for key specifications such as XHTML 1.1, CSS 2.1, SVG1.1 and MathML (all versions). None of these standards have 100% compliant products in the market nor the commitment of industry to actually provide such an implementation. The mathml implementations mostly have the status of prototypes. SVG is partially supported in some products (e.g. adobes plugin which nobody seems to use, mozilla which you have to recompile to get a partial implementation and a bunch of vector graphics packages which use it as an exchange format with other software). CSS2.1 is partially implemented by some browsers. Most of these implementations have known flaws (including mozilla and safari) and because the standard is ambiguous there are multiple correct implementations (e.g. safari won't necessarily interpret things in the same way as mozilla). XHTML is barely used beyond the 1.0 transitional stage and you have to look for actually validating webpages with a candle. Many of these standards are superceded by new standards without a single compliant implementation of the old standard on the market.

Of course there are a few succesful standards as well. For example XML 1.0 is a successtory, as is SOAP. These standards have in common that several large industries were backing them from the beginning with implementations, support and feedback. For many other w3c standards this not true. W3C standards inflation is one of the key problems on the web. Nobody seems to care anymore about w3c press releases. A few years ago a new w3c standard would have been frontpage news on all major webpages. Now it is yet another pointless specification.

I don't criticize the technical innovation behind these w3c specicications, I just ciriticize the fact that the label "recommended standard" is awarded too easily and means nothing in daily practice. Almost none of the recommended standards released by the w3c meet my definition of a standard. A standard without compliant products (not proofs of concept but actual products) is nothing but a specification for such products. Without the commitment of key industry players to actually implement these specifications you can't even call it a recommended specification.

Re:pretty obscure standard

[wasabii]

I use VoiceXML for our IVR response system. IBM has an implementation, Cisco has an implementation. There are open source implementations. There are quite a few implementations actually, in very wide use.

Re:pretty obscure standard

[krehor]

You clearly are not familiar with the industry. There are dozens of implementations (probably about 4 dozen vendors) of VoiceXML platforms, and many tool vendors throughout the world. Many millions of phone calls are handled every year by VoiceXML-based systems. Telcos, wireless carriers, and businesses (of all sizes) are using this technology. Now that the final version of 2.0 has been published, and other dependent languages are being finalized (SRGS 1.0, SSML 1.0, etc.), interoperability between platf

Too many reccomendations....

[pragma_x]

I'm all for accessability on the web. This is good stuff IMHO, but I'm wondering where the W3C will draw the line. Seems like everything is becomming a n XML grammar these days.

I shudder to think what "XML-Reccomendation XML" will wind up like. *ick*

And look at who is lined up against them

[eltoyoboyo]

I would place all of my bets on VXML.

See who the Speech Application Language Tags (SALT) Forum Members [saltforum.org] are and judge for yourself.

This is not an endorsement of SALT nor the Microsoft Speech Tools for .NET. But you would be doing your company a disservice to ignore them.

Both Voice XML and SALT are backward steps IMHO compared to programming tools available from all the major IVR/ASR vendors. The MS .NET speech tools even look primitive compared to IVR vendor tools that were new in 1998. Certainly, no