W3C home > Mailing lists > Public > public-xg-htmlspeech@w3.org > July 2011

Re: About speech request initiation and reco element etc

From: Olli Pettay <Olli.Pettay@helsinki.fi>
Date: Wed, 06 Jul 2011 17:03:13 +0300
Message-ID: <4E146B21.2080301@helsinki.fi>
To: Satish S <satish@google.com>
CC: public-xg-htmlspeech@w3.org, Bjorn Bringert <bringert@google.com>
On 07/05/2011 10:27 PM, Satish S wrote:
> I don't expect our proposal to disallow starting speech input with a
> single click because that can happen in the trusted/whitelisted sites
> scenario.

In trusted/whitelisted scenarios user or user agent has already given 
the permission to use some service.

> But I expect we'll have wording in the API proposal to
> discourage UAs from sending audio to remote servers without explicit
> user consent in some form.

Ok, sounds like we're agreeing here.
(I hope I'm not misinterpreting "explicit user consent in some form" :)
  Clicking on a web page is not an explicit user consent.)


> Cheers
> Satish
> On Tue, Jul 5, 2011 at 4:12 PM, Olli Pettay <Olli.Pettay@helsinki.fi
> <mailto:Olli.Pettay@helsinki.fi>> wrote:
>     On 07/05/2011 04:58 PM, Satish S wrote:
>             As I said
>         "Note, I'm not against <reco>, if we can find a reasonable security
>             model when it is used. "
>             So if Google is ok that user needs to give an explicit
>         permission to
>             the page before activating speech recognition, then this one
>             problem is solved :)
>         Explicit permission comes in various forms. As we discussed
>         earlier on
>         mobile devices it could be a dedicated key, on desktop browsers
>         it could
>         be a button on the window chrome, in an enterprise environment
>         it could
>         be a set of trusted intranet pages set by the domain policy and in
>         untrusted environments it could be a drop down or an info bar or a
>         dialog asking the user for an extra click. I think we all agree
>         on these
>         models and that they get explicit permission in various forms. So I
>         think the problem is solved.
>     Ok. So let me still verify - Google does not require that the API
>     allows speech recognition to be started without getting explicit
>     permission from user? (So it is ok if the API disallows the UI Chrome
>     has now - or at least the API will require UA to ask permission at some
>     point?)
>         Is everyone else ok with a <reco> style markup element and its
>         methods
>         being the JS API for speech input? A similar model could be used
>         for TTS
>         as well so a <tts> element if visible on page could provide a UI
>         similar
>         to the <audio> tag, allowing users to play/pause/seek in the
>         audio stream.
Received on Wednesday, 6 July 2011 14:03:41 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:16:50 UTC