- From: Bjorn Bringert <bringert@google.com>
- Date: Thu, 30 Jun 2011 14:31:58 +0100
- To: "DRUTA, DAN (ATTSI)" <dd5826@att.com>
- Cc: Deborah Dahl <dahl@conversational-technologies.com>, Dan Burnett <dburnett@voxeo.com>, "Olli.Pettay@gmail.com" <Olli.Pettay@gmail.com>, Charles Hemphill <charles@everspeech.com>, Michael Bodell <mbodell@microsoft.com>, "Raj (Openstream)" <raj@openstream.com>, "public-xg-htmlspeech@w3.org" <public-xg-htmlspeech@w3.org>
- What is the use case for having a single SpeechService interface
that can be ASR and/or TTS? Why not have two simpler interfaces:
SpeechRecognitionService, SpeechSynthesisService? (did we talk about
this at the f2f?)
- Many new HTML APIs don't add methods to the window object. Instead
they provide constructors for the necessary interfaces.
- The function for getting the service needs to be asynchronous, since
it might have to make network requests.
Here's a strawman alternative that addresses the above:
interface SpeechRecognitionServiceSpec {
attribute DOMString serviceUri;
attribute sequence<DOMString> supportedLangs;
// ... other criteria
// On success, a SpeechRecognitionService is passed to successCallback
void getService(Function successCallback, Function errorCallback);
}
interface SpeechRecognitionService {
attribute DOMString serviceUri;
// ... other info about the service
}
A SpeechRecognitionService can be used to create a SpeechInputRequest,
e.g. by having a SpeechInputRequest constructor that takes a
SpeechRecognitionService.
To make simple things easy, it should also be possible to skip using
these interfaces completely, and just make a SpeechInputRequest with
no service specified, or just a service URI.
The SpeechSynthesisService would be analogous, but with some different
attributes in the spec and service objects.
/Bjorn
On Wed, Jun 29, 2011 at 5:50 PM, DRUTA, DAN (ATTSI) <dd5826@att.com> wrote:
> Team,
>
> Attached is the first attempt at the query API.
> This should allow the developer upon initialization of the speech to deterministically build a user experience that matches the service capabilities returned by the service handle.
> With this approach, developer does not disclose his/her intent (what capabilities he/she wants to use) but rather asks the service to present itself with all the features it supports.
>
> At this point I don't think we know what all those capabilities are and they should be a reflection of the protocol agreements.
>
> This is just to get started and I'm sure some of the attributes and methods will have to be revisited once we understand a bit better the overall API framework.
>
> Thanks,
> Dan
>
>
>
> Thanks,
> Dan
> __________________
> Dan Druta
> AT&T | Service Standards
> m: +1 425 381-3126
> e: dan.druta@att.com<mailto:dan.druta@att.com>
> [cid:image001.png@01CBC128.5FA40B30]
>
>
>
>
--
Bjorn Bringert
Google UK Limited, Registered Office: Belgrave House, 76 Buckingham
Palace Road, London, SW1W 9TQ
Registered in England Number: 3977902
Received on Thursday, 30 June 2011 13:32:37 UTC