W3C launch new speech synthesis standard

[greg.harvey]

Hi all,

Can't see anything about this yet, but here you go. It's also on the front page of their site. Brand new language, SSML (Speech Synthesis Markup Language), from the W3C.

http://w3.org/TR/2004/REC-speech-synthesis-20040907/

Cheers,

G

[bradley317]

<robotvoice>Groovy</robotvoice>

[fatnewt]

<voice source="southpark" character="cartman">Sweeeeeet</voice>

[greg.harvey]

I'm surrounded by idiots!

[megamanXplosion]

Unfortunately, I do not see this as a good thing. CSS3 is bringing us the ability to use Aural stylesheets to basically synthesize our documents inline (and you could provide alternative stylesheets so people can choose which voice they want to hear), I do not see the point in an additional standard for synthesizing documents online. This will only lead to confusion and increased maintenance work for the majority of those trying to synthesize their pages. The SSML standard will likely hurt the Internet more than it will benefit it.

Just my two cents.

[vgarcia]

Unfortunately, I do not see this as a good thing. CSS3 is bringing us the ability to use Aural stylesheets to basically synthesize our documents inline (and you could provide alternative stylesheets so people can choose which voice they want to hear), I do not see the point in an additional standard for synthesizing documents online. This will only lead to confusion and increased maintenance work for the majority of those trying to synthesize their pages. The SSML standard will likely hurt the Internet more than it will benefit it.

Just my two cents.

SSML is meant to work with CSS/XHTML rather than instead of it:

The intended use of SSML is to improve the quality of synthesized content. Different markup elements impact different stages of the synthesis process (see Section 1.2). The markup may be produced either automatically, for instance via XSLT or CSS3 from an XHTML document, or by human authoring.

If you have aural CSS already, then chances are the SSML will be generated in the background for some device to understand and you won't have to do anything extra.

[megamanXplosion]

Why would the CSS parser generate SSML and send it to an SSML parser to append data to the DOM tree and then interact with the audio synthesis? That makes no sense at all. It would make more sense to have the CSS parser append data to the DOM tree and then have the audio synthesis occur by reading the DOM, that would be more logical, more scalable, more web-application-friendly, and less CPU intensive. Having the browser jump through a million loops and tie together thousands of proverbial strings to generate working documents is silly, which is the main reason I believe that message you quoted was nothing more than a "you could..." message, rather than a "you should..." message. I highly doubt A-CSS will map to SSML form, that would be pointless.

I really don't see the point in "human authoring" either, since A-CSS will do the job for you, and you can effectively mix up your documents with other devices easier and allow the user to have a choice on how a document should be read back to them by providing alternative stylesheets. People can already author their own audio synthesis playback with A-CSS. Training people a new language, and tell them to craft multiple documents for the same information, and maintaining two seperate copies of the document in the future, would be a complete waste of time and money.