RE: Speech API Status

Thanks Eitan,

We are thinking about voice properties beyond just gender, though there is a specific need to control gender.  As agents become more common, we think apps will care about additional voice characteristics like age, inflection, (e.g. “robotic”) and more.  The spec could be positioned to add those if voice properties were provided with the list of voices returned in response to getVoices().  And it could support gender only for now.

I came across this picture emphasizing gender (in the voice name) for speech synthesis.  Having a solution beyond adding to the name string seems preferred.

[http://media.npr.org/assets/img/2013/03/10/synthvoice071-85a00b30a117d0ab3ada1feb9d6837c064129694-s700-c85.jpg]

Jerry

From: Eitan Isaacson [mailto:eisaacson@mozilla.com]
Sent: Tuesday, July 25, 2017 2:04 PM
To: Jerry Smith (WPT) <jdsmith@microsoft.com>
Cc: Glen Shires <gshires@google.com>; Dominic Mazzoni <dmazzoni@google.com>; public-speech-api@w3.org
Subject: Re: Speech API Status

Hi Jerry!

Thanks for reaching out. The previous length parameter initiative was awesome and I'm happy we have that now.

I'm busy on other browser work and won't have many free cycles these coming months to work on new features.

I think there is important work to be done regarding this spec, that may require some big changes:

  1.  Standardize concurrent speech behavior (my pet peeve), https://www.w3.org/Bugs/Public/show_bug.cgi?id=21110

  2.  Minimize fingerprinting potential https://www.w3.org/Bugs/Public/show_bug.cgi?id=29350

  3.  An async getVoices function. The current getVoices/onvoiceschanged pattern is not ideal
As for voice gender, I feel mixed about that.

Cheers,
 Eitan.


On Tue, Jul 25, 2017 at 1:29 PM, Jerry Smith (WPT) <jdsmith@microsoft.com<mailto:jdsmith@microsoft.com>> wrote:
Glen, Eitan and others,

I see the Speech API spec is at “Unofficial Draft” status and was last edited in June of 2014.  I’ve received a few internal inquiries from groups interested in building on this API, and at least one request for a feature addition (voice gender detection).  What is the potential for moving the spec to higher maturity?  Or is the unofficial draft believed to be sufficient for the level of adoption so far?

https://dvcs.w3.org/hg/speech-api/raw-file/tip/webspeechapi.html


Regards,

Jerry Smith
Microsoft