googlesystem.blogspot.com/2008 ...
eg: knol.google.com/k/ari-green/mu ...

The "Listen" option is enabled for the Multiple Sclerosis knol:
knol.google.com/k/ari-green/mu ...

An interesting thing about that knol is that you can click "Share" and see the number of page views at the bottom left of the popup.

i34.tinypic.com/i36q7c.png

In other knols, this same message is displayed but the view count always seems to be zero.

I windier if knol in the future will do solely video knols of seminars like TED or something... Just a though.

I didn't mean it would host TED videos, but that it would do something similar to TED.com. Also I meant wonder not windier :D

This indicates that Google's text-to-speech engine understand the semantic of sentences and paragraphs better other engines, thanks to Google enormous database and collective intelligence.

I am thinking whether Google may provide such function as a service to other web pages, particularly Wikipedia. Sometimes I just want to rest my eyes when reading the Web content.

Do you think Google may do this in near future?

Don't forget that Wikipedia already started audiofying content through user-submitted audio readings. It's the usual inefficient way of Wikipedia that relies on great amounts of data.

It is great that google has started on that road. At www.dixerit.com we are speech-enabling hundreds of sites and can cover more than 20 languages, offering high quality audios, just as easily.
For example, check red.es or ticinoonline.ch for Spanish and Italian.

(Just to add, Google confirmed it's "automated text-to-speech". But they don't "have any additional details or future plans to share at this time".)

Hmm, this "Google" voice doesn't sound to me like a big improvement over other TTS engines. In fact it has the same pronunciation problems. For example listen to the word neurological in the beginning of Multiple Sclerosis article and compare to Mike voice at AT&T natural voice demo page (google for it).

I don't really understand why they offered this new service. What you can do with knol, the same thing you can do with blogger. And i am suspicious many people will paste the contents of their blogs as knol articles to create a lot of duplicates.

At ReadSpeaker we paved the way to the audio web back in1999 and have over 2000 web sites using our web based speech-enabling technology. You can listen to our service by going to the International Herald Tribune for example (www.iht.com). We cover more than 20 languages ranging from UK and US English to French, German, Japanese, Spanish, etc. You can check us out at www.voice-corp.com.

Yeah, I'v seen such service on the City of San Francisco website (www.sfgov.org). I think that is readspeaker as well. Pretty awesome service!

For those of you who forgot, Google said that this was the point of their GOOG-411 service. This TTS engine sounds exactly like the one provided by VoiceXML, the language that goog411 (sounds like) it was written in. The only difference is that this one is reading bigger chunks of text. Google's FAQ for goog411 i believe still says that the point of the ad- free service is to train their speech recognition technology. It also provided a good testing ground for TTS, I guess.

I think Google seems to have some big plans for speech, with does fit with their goal of making the world's information available to everyone, even the blind. Your suggestions seem like very strong possibilities.

Ahron, but Goog-411 was to train their speech-to-text (e.g. video search), not their text-to-speech (like Knol), Google said*... or do you think it's also useful for the latter?

* blogoscoped.com/archive/2007-1 ...

They also had to read names (was their a details option too?) of businesses to users, and not completely make foreign names incomprehensible. Google seems to really want to do something with voice. Maybe include Grandcentral for something?

The quality is nice, I think and I believe that people outside of major English countries are more satisfied :)