Re: Organized first draft of Use Case and Requirements Document

If we're talking opinions - I don't see how a keyboard and mouse (which you
buy in a store, or come with your computer) are remotely like tweaking a
setting buried in a browser control panel. To me it seems more like changing
the javascript engine your browser uses - you say it would be weird if one
website worked fine in my browser but another one didn't - yet that happens
all the time because of different standards interpretations or completeness
or quality of implementations.

When Chrome doesn't work with a website, and I have to switch to Safari, I
don't blame myself and think I should just change a setting somewhere, I
blame the developer and wait for them to fix it. We need to give the
developer the ability to select the recognizer so they can be responsible
for making sure their apps work across all browsers, and not put this
responsibility in the hands of users.

You say users need consistency - I agree. The primary "consistency" they
need is for their apps to work, not for recognition accuracy to be
identical. In any case, the same recognizer with two different grammars will
have different accuracy anyway - so it's really impossible for us to ensure
that one website understands you perfectly while another does not. I think
you may be considering this a 'dictation' only affair (in which case I do
agree with you that the same text should work in the same way given
identical conditions), but we are allowing for developer specified grammars,
for which is much harder to ensure consistency - that will be the
developer's job, hence we must give them the tools.

-Andy


> From: Satish Sampath <satish@google.com>
> Date: Fri, 8 Oct 2010 10:24:02 +0100
> To: Andy Mauro <Andy.Mauro@nuance.com>
> Cc: Michael Bodell <mbodell@microsoft.com>, "public-xg-htmlspeech@w3.org"
> <public-xg-htmlspeech@w3.org>
> Subject: Re: Organized first draft of Use Case and Requirements Document
> 
>> That said, Iım not sure I agree with having the recognizer/speech-synthesis
>> be a browser setting. A browser setting is a user control, and users arenıt
>> going to care about the recognizer/speech-sythesis they use, theyıre just
>> going to expect speech features to work. However, certain types of
>> developers are going to care about the recognizer/speech-synthesis they use,
>> and as such it makes sense for this to be an (optional) facet of the
>> markup/language.
> 
> I think users will care about speech recognition the same way they
> care about having a good input device/keyboard, installing the right
> voice recognition software or selecting the right operating system
> based on their needs. And users need consistency in terms of
> recognition for every website, it would be weird to have one website
> understand my voice perfectly fine and another to have problems with
> the same text.
> 
> --
> Cheers
> Satish

Received on Wednesday, 3 November 2010 14:16:41 UTC