Re:Parents of children called Alexa challenge Amazon
by necovek on Friday July 02, 2021 @04:29AM
(#27709101)
"Speech recognition software" responds in a very humanly voice. It's the very reason people want software assistants to have human voices, even female voices, because it makes them feel more at ease. And we don't have those sat-nav assistants saying things like "you made a mistake and missed your turn", even though we know they are robots and that might be a fact. So we are making computers behave like humans, but you then need to know the difference... how?
[ Parent
| on HN ]
Re:Parents of children called Alexa challenge Amazon
by 908B64B197 on Friday July 02, 2021 @02:07PM
(#27714751)
If there was an option to have an air traffic controller's voice for turn by turn navigation, I would enable that.
Re:Parents of children called Alexa challenge Amazon
by asteroidbelt on Friday July 02, 2021 @04:47AM
(#27709211)
> "Speech recognition software" responds in a very humanly voice.
Speech recognition software responds to any voice it was trained to. It will respond to dog barking or morse code it there will be a demand.
It responds to "humanly" speech, but it likely mostly ignores the tone. (Although, I imagine eventually Alexa will learn to start different music depending on the tone of the person asking to play the music.)
> It's the very reason people want software assistants to have human voices, even female voices, because it makes them feel more at ease.
No, it's different reason. Human to assistant and assistant to human is not symmetrical.
It's hard to retrain humans to respond to a different form of speech, but it's easy to retrain the model to whatever is needed/used by users.