Hacker News Reader

Comment by paul7986 | Hacker News Reader

A bit off topic but anyone here have access to Chat GPT voice conversations (how is it)? They said they are rolling it out within the next two weeks for plus users (which I am), yet as of now I don't see the option under "New Features."

Ever since seeing this video from last year of a journalist having a conversation with Chat GPT https://www.youtube.com/watch?v=GYeJC31JcM0&t=563s Ive been looking forward to using it (heavy Siri user).

Mix Chat GPT Voice Conversation's with Zuckerberg's new avatars (https://twitter.com/lexfridman/status/1707453830344868204) and those once in your life can still be (loved one who passed to an ex to Taylor swift.. creepy i think so but looks like that's where we are headed).

View on HN

> video from last year of a journalist having a conversation with Chat GPT

Interesting, but that's just speech recognition, ChatGPT and speech synthesis.

I'm really waiting for them to do a full end-to-end model which would allow you to have a real conversion, for example you could interrupt it. That will be crazy. It'll probably allow better speech recognition and far more realistic speech synthesis too because the information doesn't have to go through the highly lossy medium of text.

Also how did OpenAI use such a bad speech synthesis system?

Recommend pi [0] on iOS or iPad if you want to try really pretty convincing conversational voice AI.

[0] https://pi.ai/

CallAnnie is another one if you have a recent iphone.

Why does it need to be a recent one?

CallAnnie seems to use iOS built-in "neural"/AI functionality/API, which does processing on-device (rather than in-the-cloud), which requires your phone's SoC to have the gubbins for accelerating those computations (basically a mini-TPU): https://machinelearning.apple.com/research/neural-engine-tra... - Apple says you need an A11+ for iPhone (since iPhone X), or A12+ for iPad, and M1+ for macOS.

Oh thanks that makes sense. Cool that on-device capability can be leveraged

> You agree to receive automated text messages from us, even if your mobile number is listed on any do-not-call registry.

From their signup page. I'm not sure that's legal in a lot of jurisdictions you can use pi.ai from.

I just looked and I saw it available under “New Features” in their iOS App.

They really do not do a good job letting you know when features go live.

At first blush Pi.ai seems like a “better” conversationalist.

Been using PI past few hours while driving .. u can almost get lost in the conversation... it's response time though is not like a real conversation as it has to send your audio file question or response then u have to wait for it's response. Other then that it's amazing and Apple should buy them quickly a d integrate the tech into Siri asap.

Man that video of AI talking was boring. It needs to learn how to answer questions without lecturing. The tedium in its replies causes me to tune out pretty quickly.

Oh I like how it's neutral and stays away my from any bias like Pi.AI does. I asked PI many controversially biased topics like it's thoughts on Trump, trans ppl, what media outlets aren't biased and did it recommend any replying Reuters and the AP which I agree..it then asked me what news outlets I thought weren't biased and I jokingly said MSNBC and Fox News which is corrected me saying umm those are biased outlets lol

If you prompt it correctly it does.

https://news.ycombinator.com/item?id=37675274

I'm also a plus user, and heck i don't even see New Features on the iOS app lol. Where is that?

It doesn't show up for me either. The web doesn't show the Bing Browsing feature either even after multiple articles about how that was re-enabled for all Plus users. It's really annoying.

Agreed. Though i do have Bing in the web UI, but i don't have any of those features in the iOS app hah.

It's available for me on the iPhone app. The voice is really good and the pauses are very human

Reply on news.ycombinator.com