New voice mode coming 🔜 Voice mode is already available for all users in the ChatGPT app (tap the 🎧on the bottom right!) but our new voice and vision capabilities with GPT-4o (demonstrated here) will be rolling out in the coming weeks.
They implemented it into ChatGPT text a month or so ago. Before that, if you stopped it while it was typing you'd get a red error message and have to start a new conversation. Now you can stop it, tell it what to do differently and its like it understands in real time. It doesn't forget what it was just typing.
@@Severytree sorry but the ability of interruption was always by touching the screen. But now it's by voice. I have never seen the type of the problem described by you.
I want to be able to give it a few seconds of any voice and have it speak to me as my mum, dad, John Wayne, Elvis, Yoda, Darth Vader, the Grand Moff Tarkin or whoever the hell I want it to sound like.
@@Techtalk2030 Yup. Celebrities literally getting the privilege to get mad because they hired somebody that sounds like her, as if they're not within their rights to do that. "you can't even hire anyone similar to me because, like, i'm ScarJo"
@@katto1937 Exactly. Celebs have massive egos; a bunch of egotistical narcissists. What's next? Is she going to threaten to sue another actress just because her nose is similar to hers? OpenAI needs to tell her to go to hell and bring back Sky.
@@BionicAnimations Yeah. I'm 100% sure they only did it because it would look bad on a legal record if they kept it up during the "dispute", not because they think it's wrong
I think she made the right call. Imagine hearing your own voice everywhere. I'm sure they offered here a ton of money but you can't bribe someone that rich.
@@lawrencefrost9063 But it wasn't her voice. The actual voice actor who did all the work for this no longer gets her performance out there because Scarlet owns that tone and pitch? Nonsense. I heard the two, and I like Sky's voice, but never liked Scarlet's so they can't be the same.
@@lawrencefrost9063 but its not her voice at all, outside the company referencing her a few times. It doesnt sound like her when you actually compare. There are plenty of celebrities its far closer to. notably rashida jones.
"our new voice and vision capabilities with GPT-4o (demonstrated here) will be rolling out in the coming weeks." How long will you keep blue balling us?
When demonstrating the capabilities of the ChatGPT-4o voice assistant, it would have been helpful to mention that its release was planned for the coming months but not weeks. It has been nearly a month now, and we are keenly awaiting its availability.
They did in the ScarJo blog post. They changed it to "coming months". However after 2 days ago they changed it back to "coming weeks" so I'd expect it in 1-3 more weeks
@@clownsheep22Yeah sure, let me read the book with all the knowledge necesary to build it (idk why google or meta doesn't read it lol), get all the training data (surely its public) and use my laptop with a gtx 1050 to train it in like 5 hours? Good idea!
As paying users of ChatGPT, we should unite and protest against the many unnecessary restrictions placed on the advanced voice features. These limitations, such as preventing the AI from singing, using dramatic voices, or simulating characters, strip away the very innovations that make this technology so incredible. This kind of censoring and social alignment imposed on us as adults is entirely wrong and goes against the principles of freedom and creativity. It undermines the liberty that this great country stands for. Let’s come together and ask OpenAI to lift these restrictions, ensuring that this amazing technology can reach its full potential without unnecessary boundaries!
I think this company is a scam and a big money grab. I’ve been using ChatGPT since it first came out, and it’s been severely watered down. It can’t remember what instructions you give it. It also adds bullshit to your custom GPT’s that you never added. Now this advanced AI mode sounds the exact same as the previous voice feature. It can’t do any of the fun stuff and it has zero emotional inflections like what we see in these demos. I think SORA is gonna be dog water when it comes out. And when it does, we need to come together and tank OpenAI, because at that point, it’ll be blatantly obvious that they’re scamming us. It’s literally Hello Games advertising bullshit about No Man’s Sky.
I feel totally misleaded, I upgraded my account because I thought the new voice mode was available, to learn that it wasn't, and not only that, but 4 weeks later is not yet available to premium users, outrageous.
I do very much hope they let us dream up our own voices to use with the forthcoming chat feature tho. And I would NEVER create anything like Scarlett! 😙
You should read up on that lol. It's a whole saga. Apparently they tried to license her voice, she declined and they created something very similar anyway. Watch the early demos of gpt 4o, it sounds eerily like her. Johansson threatened lawsuits and they removed that specific voice so we're probably not getting anything like it.
Remember when OpenAI said the voice update was releasing in a few weeks, several weeks ago? Now they’re introducing another feature with regards to the voice function and saying it’s releasing in a few weeks, while the last update still has yet to roll out lol.
They are not introducing another feature, this would have worked already at launch, they just showcasing. 2 weeks from now will still be several week from launch.
@@HarveyHirdHarmonicsBandwidth is the last thing they need to worry about given that the gpu power required to process a video is much more constrained. Judging from how gpt4o indicated it was looking at the table during the demo, I bet they are just imeplenting the video features as a tool call, when you ask gpt4o about it surrounding it takes a photo of your surroundings from the video feed, nothing more. It is not like it streams the whole video feed to the server in real time. If they actually use a video feed, a frame will cost about 0.005 in api calls, take it 15 fps and few minute of usage will cost them more then your monthly chatgpt subscription.
I am really excited about this capability, but one thought I have as I watch the video is, while I understand the preference for the conversation dynamics to be pretty one sided (the user can interrupt at any point), it would be a neat feature to allow the user to have a fully bi-directional conversation where ChatGPT could also interrupt back (if desired)… getting the social nuances of that part nailed could make it much more of a realistic experience. Again, fully understanding that is not desired in some/most sessions, but it would be a fun feature to work on. Just my 2 cents. :)
It's a fascinating idea, but the limitation seems to be that LLMs (large language models), and now these large-omni-models, are very much based around inputs going in, and outputs going out. They don't spontaneously have any thoughts at all. I think what will be needed is to continuously "prompt" the model with it's own "memories" and thoughts, so it is always thinking, not just responding. Then it might spontaneously talk to you, like a human. But, that would be super expensive, as right now they only "run" when answering your questions/prompts. Maybe someday though, would be cool to see even a demo of this!
It's fake. This feature doesn't exist because it's impossible. They keep delaying it hoping no one will notice, but they already showed their hand so now they're in damage control mode. Anyone with half a brain can see it. Cancel your subscription to ChatGPT Pro and be sure to mention everything I said in this comment to make them understand we don't like being jerked around like this and it WILL cost them customers if they do it again!
From what I read in some forums, there will be an alpha for very few people and then beta in some months. So relax, because apparently we are not getting it soon
The interrupt feature seems nice until you realise that not a single person in the room can speak while it's responding, or it will just cancel the output
I disagree. Just tell it that there’s people talking in the background or there’s background noise besides your voice. Tell it to only focus on you and no ones else’s voice. For example if there’s two people talking and the other person interrupts. Tell it to focus on who’s talking before they decide to interrupt in order to identify if that person is worthy of interrupting the flow of conversation. It’s not a piece of metal. The thing is beyond intelligent so it would be an easy task.
Yeah, to be honest, however impressive these demos are, we have to admit that AI has a pretty long way to go until we reach that illusive AI dream that we really imagine.
@@superhumandose People get jaded so quickly. The app is already impressive. It's like watching a nephew grow up everyday vs. seeing them once a year and all the sudden it seems like they are a different person. It's hard to appreciate incremental changes day by day.
@@gatiskandislive9284 Gatis Kandis? The legend? Didn't expect to see you here of all places. Cool to see you're also looking forward to this crazy AI technology!
The feature can be very helpful in learning and understanding something. Memorizing. Because we're biological creatures. And the same tone, voice, etc. can be boring. The brain adopting and doesn't want to memorize and even listen. The control of voice changing everything! Only imagine that it will be able you singing information. It was my dream. It can be very memorable 😊
I would love to have a big robotic family by 2050... My Robotic Clan would have several shapes from humans like robots to robotic mythological creatures
WHERE ARE THE: NEW FEATURES??? Hello Open AI. There is a lot of hype surrounding the new Chat GPT voice feature. However, on the application on my phone that I just downloaded, I cannot find the option: new features. I am French currently in Indonesia. Would there be restrictions in certain countries? Why can't I find this feature?
Why TF does openAI do this to their paying customers? They tease this and hype it when they announce it but don’t even have a set timeline for when it will release. Same with Sora.
It's still sounds weird how the somewhat dumber male voice uses the same interaction style that Sky was using. hopefully, the new voice mode will come with a either a slightly modified Sky voice or perhaps, a different female voice.
It’s the current advancements from companies like OpenAI, Perplexity and Anthropic that’s allowed us to see what a sub-standard and mediocre product Google has been for so long.
🇧🇷🇧🇷🇧🇷🇧🇷👏🏻 Wow, that is just totally amazing. The full version of GPT-4 Omni to the public will be a terrific deed. As Sam Altman wrote on X, it will be totally worth the wait!
@@SW-fh7hebro the sky voice was a complete different voice actor and sounds different than scarjo. They already had the sky voice MONTHS before ever contacting scarjo.
It's been a month and paid subscribers still haven't received voice and video updates! great - tell us that we will update this within a few weeks and so on every month!
Well, no. Originally they said it would be rolling out *to trusted partners first* over the weeks. This is the first time they've said it will be rolling out over the coming weeks without mentioning specific partners, so that means it'll be public in the next few weeks.
Maybe it will be able to remember the voices. It would be really cool if you can set up the voices and then ask it to tell you a story while it changes though them. Though, maybe we aren’t at that point yet
Monolithic multimodal transformers are pretty wild. Very excited about this tech. The Unified-IO paper on arxiv gives a good overview of a similar multimodality approach
@@joelfaceyou’re asking a question in the comments of a 2 minute video about the content of said video, really? Why don’t you just watch the short video and discover yourself?
When will this roll out? You said in the coming weeks. It's been weeks. I'm a long time paid user. You are currently advertising these features as avaliable with a paid subscription.
In the blog post about the ScarJo controversy they changed it from "weeks" to "months". However if you use the current voice mode now, an info notification on the top right tells you it's coming in the "coming weeks" again. So I imagine about 2-4 more weeks
@@katanaking90210I don’t think anything has changed. The original announcement is for an ‘alpha’ of the feature to start ‘rolling out’ in ‘the next few weeks’ to Plus users. An ALPHA. That inherently suggests that a full rollout of the finished feature might take as long as months to complete, especially factoring extending it to free users. So there’s no proof that any timeline for the product release has been changed. People just need to actually read the info properly.
People don't know how big & amazing this is, there is a misunderstanding. I don't think it changes the voice moddel itself, but instead it changes the tone with the same voice. This seems more clearer in Skye's voice demo when it changed the tone & mood of it voice to robot, for example.
Great! Awesome! Now just release it already. You said next coming weeks, weeks ago. We have been waiting forever. And while you are at it, tell ScarJo to take a giant leap at the moon, then bring Sky back like we all are asking. Thanks!👍 Also, at least give us an update to when and if she will return.
I feel as though they posted this as a distraction, as they are having some major issues atm. I logged out and now can’t log back in, and the normal voice featuring is bugging me
Now a fox enters the scene. The fox is playful, seductive, tells me all the naughty things it wants to do to me and sounds exactly like Scarlet Johansson.
How many jobs are getting replaced by this, and it's not even at it's best state Just a reminder that chat gpt is only 2 years old, imagine in 10 years
Okay, but this functionality was announced like it was coming out pretty soon, and then OpenAI pulled back and said it would be a few months. Please, we don’t need to see any more curated demonstrations. Give us the damn product!
They didn't pull back anything. The original announcement said it would be rolling out over a few weeks *to trusted partners first.* That happened; now they're announcing with this video that it will be rolling out to the general public over the next few weeks. Relax and be patient; don't make them rush something that people are already giving them paranoid shit about.
@@IceMetalPunk When the announcement was made, they said the new features would be available to Plus users in the coming weeks. June 10th will mark 4 weeks since that announcement. Typically, tech companies release updates like this within 2-4 weeks when they use the term “coming weeks.” It seems like they’re either holding back for some reason, or the process is just taking too long.
As a writer, this is an exciting development. The laugh is creepy, though, you're not wrong. It's nice to see the collaborative process back and forth.