Re: [HTML Speech] Let's get started!

On 09/08/2010 05:26 PM, Satish Sampath wrote:
> Hi Olli,
>
> Adding a speech attribute to the input tag will enable web developers
> use existing form fields for receiving user input either via speech or
> keyboard/other means. The core use cases are for a text input field,
> both single line (for e.g. search box, email subject...) and multiline
> text areas, content editable elements (comments/blog posts, email
> body, ..). We have tried to mention all such controls/elements which
> currently allow text input in the proposal.


HTMLSelectElement doesn't allow text input.
<input type="radio> is in a way pretty much the same as <select>.
checkbox could be used for cases when the speech input is
"yes" or "no".
contentEditable is mentioned as a future work.

But still, because different elements need rather different handling,
I really wouldn't like to go the "speech" attribute way.
Something generic and API-wise consistent would be better, IMHO.
(and also based on my implementation experience on various ways
  to handle speech input/output in web context.)

-Olli


>
> I think it is reasonable to consider not adding the speech attribute
> for non-text input fields such as date, calendar, numbers, check
> boxes, file etc.
>
> Cheers
> Satish
>
> On Tue, Sep 7, 2010 at 6:13 PM, Olli Pettay<Olli.Pettay@helsinki.fi>  wrote:
>> Hi all,
>>
>>
>> On 09/06/2010 10:48 PM, Satish Sampath wrote:
>>>
>>> Thanks for getting us started Dan.
>>>
>>> Some of us at Google have been working on a simple API for speech
>>> recognition in HTML by extending editable HTML elements with a
>>> 'speech' attribute. A working draft is available at
>>> https://docs.google.com/View?id=dcfg79pz_5dhnp23f5 with the
>>> requirements, use cases and the API proposal.
>>
>> I was somewhat positive to the original proposal when there was just
>> simple speech input element. But the newer proposal adds speech attribute to
>> many (somewhat random) form elements.
>> And yet it doesn't handle few
>> rather basic use cases like link activation.
>>
>> I think we don't want to start adding "speech" to all sorts of
>> elements. Different elements need different speech recognition result
>> handling.
>> X+V is kind of an example when special casing elements
>> starts to make the "API" (X+V doesn't really have an API) awkward.
>> Same could be said about multimodal CSS.
>>
>> So I think we should have something closer to "simplified" SALT;
>> simple API to control ASR and TTS.
>> Even if the first version would support only ASR, we must keep TTS
>> handling in mind all the time.
>>
>>
>>   We brought it up in the
>>>
>>> WHATWG lists a few months ago and saw some positive interest, feedback
>>> from which have been incorporated into the above proposal
>>> (http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2010-May/026338.html
>>> and
>>> http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2010-June/026747.html).
>>> However it is very much a work in progress and hopefully will provide
>>> a good starting point for discussions.
>>>
>>> In order to experiment with the API and get web developer feedback, we
>>> are also currently adding the core features of this proposal to
>>> Chromium and WebKit.
>>
>> Hopefully you prefix all the methods and events with chromium or webkit ;)
>>
>>> This can be tested with the latest nightly build
>>> of Google Chrome at http://tools.google.com/dlpage/chromesxs (for
>>> windows) and will be available in the upcoming developer release as
>>> well. We already see a few web developers creating web pages
>>> showcasing the feature (for e.g.
>>> http://www.jeremyselier.com/entry/speech-attribute-demo) and hope to
>>> use it as a channel for feedback as we implement the XG's proposal in
>>> future.
>>>
>>
>>
>> br,
>>
>> Olli
>>>
>>> Cheers
>>> Satish
>>>
>>>
>>>
>>
>>
>
>

Received on Wednesday, 8 September 2010 14:44:29 UTC