- From: Daniel Weck <daniel.weck@gmail.com>
- Date: Wed, 6 Jul 2011 17:32:04 +0100
- To: "www-style@w3.org style" <www-style@w3.org>, Andrew Thompson <lordpixel@mac.com>, fantasai <fantasai.lists@inkedblade.net>
That's also how I interpreted the SSML spec. /Daniel On 1 Jul 2011, at 13:46, Andrew Thompson wrote: > On Jun 30, 2011, at 6:01 PM, fantasai > <fantasai.lists@inkedblade.net> wrote: >> No, I mean the <integer>? <gender> <integer>? notation (which >> represents <age> <gender> <variant>), as opposed to the <string>/ >> <ident> >> notation which specifies a voice by name. >> > > > Thanks. I think a use case for multiple ic voice families can be > something like this: > > voice-family: 60 male, 60 female; > > For a given language there may not be any male voices so the second > one would be used. > > Since we explicitly leave the matching algorithm up to the > synthesizer it might even be necessary to write something like: > > voice-family: 60 male, 60 female, 30 male, 30 female; > > The intent here is to prefer a mature male voice if one is > available, with a mature female next, followed by an adult male > etcetera. Essentially trying prioritize age over gender.
Received on Wednesday, 6 July 2011 16:32:34 UTC