Re: [css3-speech] “neutral” voice

Note that SSML 1.1 doesn't define "neutral" either:

http://www.w3.org/TR/speech-synthesis11/#edef_voice

I have always interpreted the 'neutral' value as "if possible, pick a voice that doesn't sound male or female in an obvious manner. Note that this may select a more robotic voice, or for example a human-sounding stylised voice".

Daniel

On 24 Jul 2012, at 01:51, Jon Rimmer wrote:

> On 23 July 2012 14:56, Jens O. Meiert <jens@meiert.com> wrote:
>> 
>> What is the definition of a “neutral” voice (family)? It seems the
>> spec so far fails to define this [1]?
>> 
>> 
>> [1] http://www.w3.org/TR/css3-speech/#voice-props-voice-family
> 
> 
> Well, the note accompanying <gender> says...
> 
>    "Note that the interpretation of the relationship between a
> person's age or gender, and a recognizable
>    type of voice, cannot realistically be defined in a universal
> manner as it effectively depends on numerous
>    criteria (cultural, linguistic, biological, etc.)..."
> 
> So, perhaps it's not possible to really define it further, as it's
> only intended as a hint whose meaning is left up to the speech
> synthesis engine?
> 
> However, a question worth addressing might be: Is "neutral" intended
> to communicate, "I don't care about the gender, use whatever is the
> cultural default", or "I do care about the gender, and I want the
> voice to be gender non-specific if possible." If it were the former,
> the speech engine would default to a male voice, but if it were the
> latter, it would use a distorted or robotic voice.
> 
> My interpretation would be that it is the latter, but perhaps it
> should be made clear.
> 
> Jon
> 

Received on Tuesday, 24 July 2012 08:21:15 UTC