- From: Dan Burnett <dburnett@voxeo.com>
- Date: Fri, 29 Jul 2011 17:16:50 -0400
- To: public-xg-htmlspeech@w3.org
Group,
The minutes from the last call are available at http://www.w3.org/2011/07/28-htmlspeech-minutes.html.
For convenience, a text version is embedded below.
Thanks to Milan Young for taking the minutes.
-- dan
**********************************************************************************
HTML Speech Incubator Group Teleconference
28 Jul 2011
[2]Agenda
[2] http://lists.w3.org/Archives/Public/public-xg-htmlspeech/2011Jul/0028.html
See also: [3]IRC log
[3] http://www.w3.org/2011/07/28-htmlspeech-irc
Attendees
Present
Daniel_Burnett, Robert_Brown, Milan_Young, Michael_Johnston,
Patrick_Ehlen, Debbie_Dahl, Olli_Pettay, Charles_Hemphill,
Bjorn_Bringert
Regrets
Chair
Dan_Burnett
Scribe
Milan_Young
Contents
* [4]Topics
1. [5]Protocol report
2. [6]Protocol work
3. [7]Organizing parameters of a function
* [8]Summary of Action Items
_________________________________________________________
Protocol report
Robert: About 1 week behind scedule
... Draft review is public
... Expect to incorperate redraft in one week
Dan: Need to discuss timings
Robert: Seems like a logisitcal challange
Bjorn: What is use case for async sending notifications?
Robert: Still debating
Dan: Let's talk now
... maybe at end of call
... Michael is still on vacation
... but doesn't look like alot of progress on API group
Bjorn: Agree that there has been no progress
Dan: Will speak with Michael when he returns
Bjorn: There are many outstanding tasks which were assigned by
Michael
Dan: Agree that we need to follow-up on those tasks
Debbie: It's helpful if people just do their own requirements
... rather than wait for Raj
Robert: Perhaps we should do a task "role call" now
Dan: Good idea
<burn> Reminder email is
[9] http://lists.w3.org/Archives/Public/public-xg-htmlspeech/2011Jul/0018.html
Dan: No communication from Raj
Charles: Did re-review Microsoft proposal
... need to flesh out more of the JavaScript API and maybe make a
case for markup
Dan: Need to send everyting ASAP
Charles: Will try to send by next week
Dan: Don't have my task done, and no idea when it will be complete
... and no estimate
Oillie: Have something in mind, but need to formally write up
... will send in next couple days
Dan: Did recieve something from DanD, but not on the call today
... but it needs more discussion
... Did recieve something from Debbie
Debbie: Would like to review item #5
... If time today would like to discuss how to organize parameters
... could get something out by middle of next week
... don't expect a full rewrite, just cleanup and making more
specific
Dan: Want to review Bjorn's proposal
... but haven't been on the call
... Did Michael fufil his taks?
... Doesn't seem like he has
... That is complete list of assignments
... but still unassigned tasks
... can do this on email, not just by conference
... Anything that needs to be discussed before charter expires?
... Floated idea to IETF that we may want to formalize the work of
the protocol group
... Can take a few months to get a group together, so think about
this now
... Need to "socialize" and then BOF session for next IETF
(November)
... BOF writes the charter
... correction, BOF only approves the charter
... so should do this in advance
... had some conversations with Francois (from W3C)
... said we were well positioned for a working group creation
... next week is avaible for discussion for subgroup
Protocol work
<burn> milan: Bjorn did you get my email on async events?
<burn> bjorn: yes. why would there be an event rather than something
generated in UA?
<burn> milan: there are many different modalities.
<burn> bjorn: so why do we need an event channel?
<burn> bjorn: this would be adding a feature for something else to
hijack speech channel and turn it into generic notification channel.
it's not necessary.
<burn> charles: do we really need async results?
<burn> milan: you can request something from server and it might
have a delayed result
<burn> bjorn: all responses are asynchronous
<burn> charles: why do we need events separate from requests?
<burn> milan: client wouldn't request a response, but server just
sends event
<burn> bjorn: doesn't belong. all info from server may be delayed
but must still be in response to a request
<burn> milan: okay
<burn> milan: ... I'm fine with that
Organizing parameters of a function
<ddahl>
[10] http://lists.w3.org/Archives/Public/public-xg-htmlspeech/2011Jun/0086.html
Debbie: How to ballance parmeters to put in a recognition request vs
on the general channel
... Pick most important properties to put on the recognizer request
Bjorn: Everything should be an attribute on the recognition object
Dan: Discussed parameter effects being immediate
Bjorn: Might have an "updateParameter" method to for them to take
effect
Charles: Pause / resume functionality ?
Dan: Continous recognition means change at any point
<burn> bjorn: what's the use case for changing params while reco is
running?
<burn> milan: that's how client can send events to server, e.g.
mouse click occurred
<burn> bjorn: that's different from setting a param
<burn> milan: we don't have a way to send without setting parameters
<burn> bjorn: in protocol maybe, but we're talking about webapi now
<burn> robert: may be difficult to coordinate. better to send new
grammar and then say when grammar is to begin applying
<burn> ... in continuous reco may transition through many different
modes. audio needs to be continuous, need to say that grammars
changed.
<burn> ... won't have to deal with maxnbest, etc. can set that up
before reco. won't change language
Dan: Interesting mentioned language
Robert: Might have to change reconizers with the language
Bjorn: method to apply current parameters
Robjert: Consdier XHR
... if no outstanding request, then can change parameters
Bjorn: Then two types of parametesr
Charles: Runtime vs setup parameters
Dan: Always have some take place immediately and some not, but how
does that affect API?
<smaug> setParameters({ param1: value, param2: value2})
<burn> bjorn: atomicity is the issue, but nothing else in HTML has
atomicity
Robert: Example of nice HTML API
... perhaps Bjorn could send over email
Debbie: Leave this as an open issues
... will start email thread
Received on Friday, 29 July 2011 21:17:34 UTC