When do you expect official release of Voice model API for gpt4o?

Hi
Listening to voice messages and responding directly with a voice response in GPT-4o is amazing. I’m curious about when this will be officially released as an API.

4 Likes

there is the US presidential election this coming November. given the possibility that it could be used for nefarious reasons even with safeguards, i would bet that it will be released afterward.

1 Like

If they intended to release it after November from the beginning, why did they say it would roll out in the “coming weeks” back in May? It wouldve been more accurate to say “in the coming months”

4 Likes

We have had the ability to create realistic sounding deepfakes of people’s voices and even videos of them for like, at least a year, and that’s just how long it’s been easily accessible to a layman who downloaded an app. People with more knowhow have been doing this for years.

The voice model of GPT-4o is not introducing any potentially democracy destroying tools that we haven’t already had for years. Plus, they said it’s coming out in the coming weeks for alpha testers and will be rolled out in the coming months for all Plus users. 6 months is not exactly what I’d describe as “the coming months”, What are you talking about?

Well, the White House is currently whining about “cheap fakes” (which are actually real videos) circulating the internet, and I presume they’re implying that these “fake” videos (which are actually real) were made by AI. That will spook a lot of the higher-profile AI companies into being very gun-shy about releasing anything in the next six months. OpenAI has already entered a weird phase where they like to talk about all the stuff they have planned, but they don’t seem to want the general public to actually have access to it. So I wouldn’t count on actually getting anything interesting from OpenAI for a while.

This is a ridiculous response. Even after November, there will be other future elections, if this were the reason it was getting held back, it wouldn’t be released at all due to this fear.

They were likely confusing the fact that OpenAI, and others, are waiting until after the election to release their next big model. This is true, but it has nothing to do with “deep fakes”; that’s just PR nonsense. It’s because the industry doesn’t want the society altering capabilities of the next big models to become a negative talking point of the presidential election and candidates arguing over who’s going to implement tougher restrictions for obvious reasons. The voice features of 4o has seemingly been delayed, but it’s not for this reason. 4o is what they wanted to get out now since they’re waiting until after the election to release 5, or what ever they decide to call it.

Apple is probably going to get the new voice mode for their new Siri AI. They made a deal with OpenAI and now ChatGPT plus subscribers won’t get it.

1 Like

Open ai “released” the new voice feature back on May 13th. August rolls around and still no actual release, but we keep getting more videos of open ai employees playing with it. Glad they’re happy i guess.

3 Likes

For those tired of this, google actually just beat open ai to market. Check out “gemini live”, which is a new chat mode for gemini that is a direct competitor to gpt4o advanced voice. Though to be fair, it’s hard to be a competitor to something not actually released, but still. Open ai slow rolled this so long that google caught up and released first. Gg fellas.

2 Likes

is there an API yet for Gemini Live?

1 Like

how can we use voice to voice in model is there any api to it for gemini live or gpt4o. please someone mention here.