W3C home > Mailing lists > Public > www-voice@w3.org > October to December 2003

RE: Question about using TTS via the prompt element in VXML

From: Tracy Boehrer <tboehrer@calltower.com>
Date: Wed, 3 Dec 2003 13:54:39 -0800
Message-ID: <676743E57886C340A628FB1DCCF42D3B1BB389@CTMAIL.calltower.com>
To: "Roopa Trivedi" <rotrived@cisco.com>, <www-voice@w3.org>
Well, we would give those different attributes, by using the "variant" attribute of <voice>.  If the script specifies some set of attributes we don't have, we pick the closest match.  For example, we may only have "US English, Female" defined, and the script writer asks for "US English, Teenage Female".  We would pick the former.
 
I can understand where you are coming from though.  But even if the spec stated defined vendor values, you would run into a similar issue if the script was moved to a platform that didn't have that engine.  Or more likely, because of license costs, or quality, the platform itself changes vendors.  The script, of course, would have to change, or somehow the platform would have to default to something...
 
But the <voice> element has the "name" attribute: "optional attribute indicating a platform-specific voice name to speak the contained text."  But obviously, since that is platform specific, it's probably not very portable.

	-----Original Message----- 
	From: Roopa Trivedi [mailto:rotrived@cisco.com] 
	Sent: Wed 12/3/2003 1:44 PM 
	To: Tracy Boehrer; www-voice@w3.org 
	Cc: 
	Subject: RE: Question about using TTS via the prompt element in VXML
	
	

	What about the scenario where 2 vendors support the same gender, age,
	language etc and the script writer specifically wants to use vendor #1?
	Is the "name" of the voice recommended to be used as the distinguishing
	factor? Can we assume that this "name" will be unique across all
	vendors?
	
	-----Original Message-----
	From: Tracy Boehrer [mailto:tboehrer@calltower.com]
	Sent: Wednesday, December 03, 2003 1:35 PM
	To: Roopa Trivedi; www-voice@w3.org
	Subject: RE: Question about using TTS via the prompt element in VXML
	
	
	I can't speak for W3C of course, but I can't imagine how being able to
	specify a TTS vendor would be portable, or even belong in the spec.  I
	would think this is a platform specific extension.
	
	The way we handle it is our TTS server can be configured to start an
	arbitrary number of TTS engines (from various vendors).  But the script
	writer can only access them based on speech markup elements/attributes.
	Namely "voice", "gender", "age", "language", etc...  That way, if we
	change the vendor for the "Adult Female", the script writer doesn't know
	(or in most cases care).
	
	        -----Original Message-----
	        From: Roopa Trivedi [mailto:rotrived@cisco.com]
	        Sent: Wed 12/3/2003 1:22 PM
	        To: www-voice@w3.org
	        Cc:
	        Subject: Question about using TTS via the prompt element in VXML
	       
	       
	
	
	        In VXML, if we're using TTS via the prompt element, there could
	be a
	        variety of vendors a user could use to perform TTS. The VXML 2.0
	spec
	        does not provide a mechanism to specify which vendor/product to
	use. Is
	        there a recommended way to specify this? Are there plans to add
	this to
	        future VXML versions?
	       
	        Thanks,
	        Roopa
	       
	       
	
	
	

Received on Wednesday, 3 December 2003 16:56:32 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 30 October 2006 12:48:59 GMT