15:50:35 RRSAgent has joined #htmlspeech
15:50:35 logging to http://www.w3.org/2011/10/13-htmlspeech-irc
15:50:40 Zakim has joined #htmlspeech
15:50:57 trackbot, start telcon
15:51:00 RRSAgent, make logs public
15:51:03 Zakim, this will be
15:51:03 I don't understand 'this will be', trackbot
15:51:05 Meeting: HTML Speech Incubator Group Teleconference
15:51:07 Date: 13 October 2011
15:51:09 zakim, this will be htmlspeech
15:51:09 ok, burn; I see INC_(HTMLSPEECH)11:30AM scheduled to start 21 minutes ago
15:51:18 Chair: Dan_Burnett
15:51:42 Agenda: http://lists.w3.org/Archives/Public/public-xg-htmlspeech/2011Oct/0016.html
15:52:16 INC_(HTMLSPEECH)11:30AM has now started
15:52:23 +[Voxeo]
15:52:35 zakim, [Voxeo] is Dan_Burnett
15:52:35 +Dan_Burnett; got it
15:52:40 zakim, I am Dan_Burnett
15:52:40 ok, burn, I now associate you with Dan_Burnett
15:58:00 +Michael_Bodell
15:58:23 +[IPcaller]
15:58:49 Zakim, [IPcaller] is Olli_Pettay
15:58:50 +Olli_Pettay; got it
15:59:07 Zakim, nick smaug is Olli_Pettay
15:59:07 ok, smaug, I now associate you with Olli_Pettay
15:59:17 + +1.425.580.aaaa
15:59:29 zakim, aaaa is Dan_Druta
15:59:29 +Dan_Druta; got it
15:59:38 DanD has joined #htmlspeech
16:00:00 ddahl has joined #htmlspeech
16:00:06 +Debbie_Dahl
16:00:14 Zakim, why do you recognize me occasionally as [IPcaller] and occasionally as ??Px ?
16:00:14 I don't understand your question, smaug.
16:00:17 :)
16:01:24 zakim, who is here?
16:01:24 On the phone I see Dan_Burnett, Michael_Bodell, Olli_Pettay, Dan_Druta, Debbie_Dahl
16:01:29 On IRC I see ddahl, DanD, Zakim, RRSAgent, burn, smaug, trackbot
16:02:24 +Milan_Young
16:02:51 + +44.760.705.aabb
16:03:03 Milan has joined #HtmlSpeech
16:03:21 zakim, aabb is Satish_Sampath
16:03:21 +Satish_Sampath; got it
16:03:26 Charles has joined #htmlspeech
16:04:31 +Charles_Hemphill
16:05:27 Scribe: Charles_Hemphill
16:05:31 ScribeNick: Charles
16:05:37 Agenda: http://lists.w3.org/Archives/Public/public-xg-htmlspeech/2011Oct/0016.html
16:06:26 MJ has joined #htmlspeech
16:06:30 satish has joined #htmlspeech
16:06:36 Topic: Quick API review, particularly the continuous case 2. Last chance -- any conflicts between API and protocol?
16:06:50 +Michael_Johnston
16:07:24 Michael: sent out updated API
16:07:35 ... some editorial tasks remain
16:07:38 s/2. Last chance -- any conflicts between API and protocol?//
16:08:08 Updated API email is at http://lists.w3.org/Archives/Public/public-xg-htmlspeech/2011Oct/0017.html
16:08:17 ... didn't renumber yet to avoid confusion.
16:08:20 Updated API document is at http://lists.w3.org/Archives/Public/public-xg-htmlspeech/2011Oct/att-0017/speechwepapi.html
16:09:01 Michael: some minor questions
16:09:07 ... default way to get language
16:09:10 ... end envent
16:09:15 ... end event
16:09:47 Michael: 7.1 set language
16:10:05 ... default values with an attribute - what about language for the default value
16:10:11 Satish: pick up from th document
16:10:15 Michael: how to phrase?
16:10:22 +[Microsoft]
16:10:36 Satish: take language mentioned in the markup - if not markup, take off the body tag
16:10:49 ... reco element, can specify lang attribute
16:10:57 zakim, [Microsoft] is Robert_Brown
16:10:57 +Robert_Brown; got it
16:11:01 robert has joined #htmlspeech
16:11:22 Michael: user agent sets if not otherwise set?
16:11:46 Satish: can mandate something in the API - when connect to the speech service
16:11:59 ... page specified, for example.
16:12:09 + +1.408.359.aacc
16:12:20 ... optional in the API - can pick up in the UI
16:12:33 glen has joined #htmlspeech
16:13:29 zakim, aacc is Glen_Shires
16:13:29 +Glen_Shires; got it
16:13:38 Michael: When UA communicates with speech service, needs to communicat language.
16:14:06 ... can go off html body language
16:14:26 ... value assigned to that attribute in javascript?
16:14:42 ... assigned at open? Can JavaScript get it?
16:14:49 ... only for user setting?
16:15:03 ... only for communicating with the speech service.
16:15:12 Ollie: Assume null unless set.
16:15:39 ... UA would send default
16:15:58 Satish: agree with that also - gets sent in the protocol
16:16:15 Debi: Script could change?
16:16:23 Satish: yes, could change.
16:16:32 Dan: But script can't read the default value.
16:16:38 ... seems like a failure
16:16:46 Satish: quite common
16:16:54 ... e.g., width of element.
16:17:04 Dan: try to get value before set?
16:17:16 Satish: empty object or null string or empty string.
16:17:51 Dan: Go with it if that's how it works.
16:18:00 Charles: Should check that behavior.
16:18:16 Satish: can check
16:19:33 Michael: Section 7.3, end of it
16:19:58 ... from incorporating idea to move away from binding connection - 4 methods
16:20:06 ... and 3 event handlers
16:20:12 ... open start and end events.
16:20:23 ... what is end for?
16:20:33 ... disconnect as a result of abort?
16:20:47 Ollie: Need progress event
16:20:58 ... load end event after load
16:21:12 ... always an end event
16:21:31 Michael: get when end of reco is dont
16:21:37 s/dont/done
16:21:44 ... no matter what the cause.
16:22:12 Satish: lang attribute - don't assign, then get empty string. Assign and get the same value.
16:22:22 Michael: what about width?
16:22:34 Satish: Thought lang more appropriate.
16:23:39 Charles: Try a known attribute to avoid custom attribute behavior.
16:23:43 Satish: will check.
16:24:14 Michael: end event - fired at the end of the connection no matter what the cause.
16:24:30 Dan: when end of reco is done, no matter what the cause.
16:24:44 Topic: issues raised in e-mail conversations.
16:24:57 Satish: alternates
16:25:00 Topic: continuous recognition
16:25:17 Satish: what was missing in the first proprosal?
16:25:32 ... can add what was missing or add alternates in the most recent proprosal.
16:25:43 s/Topic: continuous recognition//
16:25:43 Michael: Talk about the current proprosal - missing something?
16:25:55 ... Did talk about alternates.
16:26:16 Satish: When results not finalized yet. Get when finalized? Don't see that.
16:26:23 s/issues raised in e-mail conversations./continuous recognition -- alternates/
16:26:43 Michael: Still have alternates. Contained in the triple.
16:26:54 ... have n-best alternate list, even with final results.
16:27:14 Satish: Alts can span word boundaries?
16:27:22 Michael: Yes - no word boundaries.
16:27:40 Satish: Can't have one alternate that spans more than one result?
16:27:52 Michael: Can't change the number of results.
16:28:06 ... easiest to write up - Milen had longer example.
16:28:31 ... select or highlight word of phrase - might want alternates to pop up for correction - this is supported.
16:28:39 s/Milen/Milan/
16:28:48 ... different orders would be more difficult.
16:29:14 ... to change boundaries, need giant correct
16:29:24 ... when with n-best correction mechanism.
16:29:36 Satish: OK if we have alternates for final results.
16:29:44 ... seems fine
16:29:57 ... have one example that shows alternates for final results.
16:30:10 Milan: Questions: how to represent finalized elements?
16:30:33 ... every item in the array has an attribute for final (boolean)?
16:30:40 Michael: Yes.
16:30:48 Milan: want to index from 0?
16:30:54 Michael: Yes, from 0.
16:31:02 Milan: will send out new example.
16:31:14 ... finals with boolean flag, plus alternates.
16:31:46 Dan: Ollie brought up Glens proposal.
16:31:57 ... didn't discuss so far.
16:32:05 ... can walk through?
16:34:28 Dan: reco from text string?
16:34:38 Michael: Don't have it - at least not directly.
16:34:51 ... can reco from uri - can use data uri.
16:35:01 ... not direct or straighforward.
16:35:28 Regd. reading back a lang attribute that was not yet set - I tested with the css width property and reading back a width attribute that was not set returns an empty string.
16:35:36 Dan: Obvious and simple to do or not.
16:35:47 HTML without width set: Hello
- shows a dialog with an empty string
16:35:50 Debi: want to set parameter recognized from text.
16:36:03 HTML with width set: Hello
- shows a dialog with "100px" as the text
16:36:21 Michael: May interfere.
16:36:35 Robert: Can do otherthings.
16:36:54 ... call emulation rathe than start that uses text property?
16:37:00 Dan: like that.
16:37:14 ... important for it to be obvious that coder is not using audio.
16:37:22 ... needs to be clearly different.
16:37:28 ... emulate good way to do it.
16:37:50 Debi: Don't need to worry about parameters that don't make sense such as end-point detection.
16:37:59 Dan: But doesn't hurt for them to be there.
16:38:09 Robert: Can ignore them.
16:38:23 Dan: attributes, not parameters, so it can work.
16:38:44 ... simpel to do? write up 1 new method and description -
16:38:56 ... certain parameters ignored or have certain values.
16:39:09 ... e.g., result has confidence of 100?
16:39:35 Debi: no - might be doing some parsing, although reco confidence could be 100.
16:39:47 Michael: might not be 100 even if emulated.
16:40:04 Robert: have 3 different emulation implementations - leave as open as possible.
16:40:37 Debi: looks like start method.
16:40:47 ... events that come back are the same
16:40:55 ... ones that don't make sense don't come back.
16:41:37 Dan: semantic interpretation - would rather not have onsoundstart, etc. come back.
16:41:41 ... different with emulation.
16:42:04 ... with substatial emulation, could be parameters for everything.
16:42:31 Debi: make as similar as possible and fine tune later.
16:42:44 Dan: what would be the harm if onsoundstart, etc. came back.
16:42:57 ... should know what that means if they call the emulate method.
16:43:14 ... just need to worry about confusion that it was from a start.
16:43:30 Robert: uneasy about emulation spec
16:43:49 Dan: mrcp needs semantic interpretation only.
16:44:04 Robert: not everyone wants to do tha.
16:44:21 s/tha/that/
16:44:42 Dan: one shot case should not be a problem.
16:44:58 ... what about custom pronunciations, etc.
16:45:07 ... but can be useful in common cases.
16:45:15 Robert: those are the common cases.
16:45:34 ... people choose unusual names for things.
16:45:47 Dan: if speaking, then normal recognition.
16:46:03 Robert: Simulating spoken input.
16:46:26 Milan: want all methods?
16:46:36 Robert: Don't - don't think they are meaningful.
16:47:02 ... wouldn't want to fire onaudiostart unless there is a system that can represent them.
16:47:12 Dan: Would like interpret rather than emulat.
16:47:20 s/emulat/emulate/
16:47:31 ... utf-8 text and interpret it.
16:47:50 Robert: Find semantic interpretation to be almost useless in the vast majority of apps.
16:48:09 Milan: app with SRGS grammar, want to send text to it.
16:48:20 ... doesn't help with punctuation, etc.
16:48:25 ... must match tokens.
16:48:34 Michael: emulation can do something smart.
16:49:02 Robert: can do something like that - works perfectly when testing, but confusable parts in reality with audio.
16:49:10 ... won't get n-best, etc.
16:49:32 Milan: given range of target users - worth having mechanism to send a string and get an interpretation.
16:49:49 ... experienting for own site.
16:49:58 s/Milan:/Michael:/
16:50:00 s/experienting/experimenting/
16:50:17 Michael: be careful how we describe the method.
16:50:34 ... ask for confusable parts.
16:50:42 ... be careful about what we say we get back.
16:51:07 ... Sounds like there is rough agreement for emulate or "recognize from text" method.
16:51:12 Dan: not emultate.
16:51:19 Robert: interpret.
16:51:33 Dan: clear if pass in as a parameter.
16:51:54 Michael: do want to trigger event - can get nomatch.
16:52:51 Dan: difference between result events and progress (audio related) events.
16:53:23 Dan: would rather not get audio-related events.
16:53:33 Michael: Can get other result oriented events.
16:53:40 ... oninterpret?
16:53:51 Dan: needs to be an end - maybe not oninterpret.
16:54:25 Michael: oninterpret similar to onstart - make changes while waiting for results to come back.
16:54:38 ... might change the UI while interpreting.
16:54:54 Dan; Could take time - far enough. In favor of oninterpret.
16:55:50 Satish: Don't set a width - get empty string. Set and then get what you set.
16:55:58 ... in meeting notes.
16:56:08 Michael: that approach should work for language.
16:56:13 Satsh: yes.
16:56:15 http://lists.w3.org/Archives/Public/public-xg-htmlspeech/2011Oct/0000.html
16:56:19 s/Satsh/Satish/
16:56:36 Topic: Glen's proposal.
16:56:46 above URI is for Glen's reco tag proposal
16:56:53 Glen: Be more declarative.
16:57:07 ... simple for html developers.
16:57:15 ... support copy and past.
16:57:28 s/past/paste/
16:57:46 ... keep simple things simple.
16:57:56 ... additional option as a permissions model.
16:58:18 ... icon that indicates "I'm speech enabled".
16:58:25 ... user gives permission by clicking.
16:59:41 ... make it obvious when audio is captured.
17:00:19 ... says that this is a spot where you can speak into a Web page.
17:01:00 ... example 1 - no specific binding except in JavaScript except onresult.
17:01:17 ... methods intended to match what's in JavaScript.
17:01:52 ... example 4 - does appending rather than overwriting.
17:02:16