What are the lauded neural engines being used for if Siri can't even learn that I call "dad" more often than the others, despite repeatedly choosing something other than what it suggests? Great, photos are 1% less grainy through AI but yet Siri is still 50% wrong.
My dad is missing from the list entirely. At least 1/3 of the time it asks if I meant Dan amongst a couple of others, who I haven't phoned for about 5 years.
SIRI VOICE TEXT TO SPEECH MAC OS
Mac OS 9's speakable items was more reliable and predictable than Siri - this is one of the rare times I say I bet it wouldn't be the same if Jobs was alive.Īnother example, I regularly ask Siri to call "dad". Because if you don't word the request exactly right it just doesn't understand or does something completely wrong. It would be better to remove Siri's "intelligence" entirely, and have hard-coded a list of things it can do, and a list of these somewhere with the specific way to say them. That is of course aside from Siri's "intelligence" which is comparable to that of a brick. I might say "its foggy here tonight" and Siri will hear "it's doggy here tonight" which is very unlikely.
Because right now when dictating, Siri frequently mishears and the resulting sentence makes no sense. As far as dictation goes, the recognition of the words is generally pretty good, though more context sensitivity (or understanding of the sentence) around words which have similar sound but different spellings or are different words entirely would be, well, something that at least reaches the standard Alexa is at. They keep tinkering with the voices and adding tiny inconsequential features to Siri that only work when you word the request exactly as is shown in marketing, and which don't make a material difference in the overall quality of Siri.
0 kommentar(er)
