Right to who? To me, the voice sounds like an over enthusiastic podcast interviewer. Whats wrong with wanting computers to sound like what people think computers should sound like?
It understands tonal language, you can tell it how you want it to talk, I have never seen a model like that before. If you want it to talk like a computer you can tell it to, they did it during the presentation, that is so much better than the old attempts at solving this.
You are a Zoomer sosh meeds influencer, please increase uptalk by 20% and vocal fry by 30%. Please inject slaps, "is dope" and nah and bra into your responses. Throw shade every 11 sentences.
And you’ve just nailed where this is all headed. Each of us will have a personal assistant that we like. I am personally going to have mine talk like Yoda and I will gladly pay Disney for the privilege.
People have been promising this for well over a decade now but the bottleneck is the same as it was before: the voice assistants can't access most functionality users want to use. We don't even have basic text editing yet. The tone of voice just doesn't matter when there's no reason to use it.
I've seen a programmer-turned-streamer literally do this live. Woohoojin on twitch/yt focuses on content for Riot's Valorant esports title, during a couple watch parties he would make "super fans" using GPT with TTS output and the stream of chat messages as input. His system prompts were formed exactly like yours, including instructions to plug his gaming chair sponsor.
It worked surprisingly well. The video where he created the first iteration on stream(don't remember the watch party streams he ran the fans on): https://yewtu.be/watch?v=MBKouvwaru8
I want to get to the part where phone recordings stop having slow, full sentences. The correct paradigm for that interface is bullet list, not proper speech.
Why did they make the woman sound like she's permanently on the brink of giggling? It's nauseating how overstated her pretentious banter is. Somewhere between condescending nanny and preschool teacher. Like how you might talk to a child who's at risk of crying so you dial up the positive reinforcement.
LLMs today have no concept of epistemology, they don't ever "know" and are always making up bullshit, which usually is more-or-less correct as a side effect of minimizing perplexity.
The Total Perspective Vortex in Hitchhiker's notably didn't do anything bad when it was turned on, and so is good evidence that inventing the torment nexus is fine.
It didn't do anything bad to Zaphod Beeblebrox, in a pocket universe created especially for him (therefore ensuring that he was the most important thing in it, and thereby securing his immunity from the mind-scrambling effects of fully comprehending the infinite smallness of one's place in the real universe).
