วีดีโอ

I don't think it is a fail. I think it is more that there is not a limit to how many voices it can do- other than it is copyright limitations on imitating celebrities. Think of it this way: how many people can you imitate? You won't know until you try. So, once you get access to Advanced Voice, just try it out and see how many you can do. If there are some voices you would like me to try, please list them out. I have done more videos. I just need to edit them out. I plan to keep doing them as this is as important tool. Thank you for the feedback.

I experimented the other night, by asking it to create a list of 100 different examples of diverse voice styles/accents, and it nailed every one of them. It had things like, "Sullen goth teen" and "sleazy salesman" and "cheerful carnival barker", "Noir detective", "overenthusiastic gameshow host", and "Transatlantic 1940's American stage actor," et cetera. It nailed them all. I think the reason it can't give an exact number is because some of them seem to be variations or combinations of base accents. For example, there's an English voice (including variations like Cockney or proper English), but there's also an "English Butler," which is just kind of a slightly slower, modified proper British voice. It's almost like each accent has a base accent, but it also has modifiers like speed, pitch, grammar, etc. So in other words, it seems to be capable of modifying voices into variations which play off of the base voice. As examples, you can ask it to speak with a very mild Jersey accent., or a very thick Jersey accent. You can ask it to sound out-of-breath, or "very out of breath, as if running up a steep hill." You can ask it to sound like a "Jersey girl with a very slight hint of Russian accent, who is completely out of breath after running up a very steep hill," or other nuances. I won't go so far as to say there are endless variations, but I don't think it's as simple as just having a list of a couple of hundred accents.

@@ThisEpicLife This is an excellent explanation! Thank you! Sleazy salesman. I look forward to giving that one a try.

@@ThisEpicLife The best use I heard so far had nothing to do with voices. My son and I were stuck in heavy traffic coming back from an appointment and he had a grad class in an hour. He asked Advanced Voice: "What is the difference between ROS I and ROS II?" He then grilled Advanced Voice for the next hour with follow-up questions. By the time I dropped him off at the university, he was completely prepped for his lab. As I was listening in on this conversation, not only did I learn something about Robotic Operating Systems, but I thought: "This is the way the technology should be used."

Oh, what a great tip! Thank you!

Kids love it for bedtime storytelling and it is very good for developing their verbal skills. Alex and I used it today to learn some Norweigan.

Thank you, Manuel. I appreciate the clarity, and I am very much looking forward to trying the new version. Admittedly, even in the current version, I find VoiceMode very useful. Getting rid of the latency will be nice, but I don't need any of the other features. I rarely ask my Alexa unit to whisper to me. And, I can't imagine a use case where I need ChatGPT to sing to me.

You're not wrong. :) I'll admit I was being a total Karen to the poor model while testing it.

I still don't have the singing, whispering version. But I still use ChatGPT Voice every day, sometimes multiple times a day. I refer to it as my "Oracle." I'm sure it will be even more useful with the updates, but even in the current form, I find ChatGPT Voice speeds up my workflow. For example, I brainstorm article ideas with ChatGPT Voice while walking the dog. To the other dog walkers, I probably look like I'm on the phone. But, I am actually "on the phone" with ChatGPT- getting some work done.

@@DeepLearningDaily Beware, when I use ChatGPT-4o, I have to tell it it's wrong numerous times. It agrees with me and fixes it's answers, but it is I who must know that something is wrong. In historical facts, in cultural etc.. So I'm not saying it's all bad, but I'm not saying it has no faults, either. So do a fact checking, for your own good. That is before the advanced features...

@@XRos28 I agree. Thank you for pointing this out. You have to know your subject material. I appreciate the heads-up, though. When I ask GPT4o to do research, I will often say: "Do your research" in my prompt. Before I publish anything, I run it through Perplexity and say: "Please fact-check this." (Why do I say 'please?' I don't know. I'm a crazy person.) Perplexity finds an actual source I can verify for everything Gpt4o says. I can also ask 4o to provide sources, but often, the links are fictitious. It saves time to ask Perplexity to do it.

Hi Lights and Colors- It's GPT4o, since I'm a Plus subscriber. But, they just haven't added the new functionality yet. They said the new functionality is coming for Plus subscribers in the next couple of weeks. It's been two weeks, so I've been checking every day. At one point in the video, I hold up my phone so you can see it says GPT4o on the top of the screen. If I do another video, I'll make this point a lot clearer. Thank you so much for the feedback!

@@DeepLearningDaily i see, so it's just the delay. I bet you're excited as everybody else. I am. I find it so out of this world, fiction-come-true stuff. 🥰❤

@@theeyes-fx6ld Yes! You captured it exactly!

This update is going to be one of the most advanced futuristic things that I have witnessed, judging by the GPT4o ChatGPT and Microsoft Copilot videos. Almost doesn't even seem real. I'll believe it when I can finally test it. I'm excited as well

This is an interesting general observation of humanity. Do you want me to pose that question to the Ember voice in my next video?

@@TOMTOM-zj5xj I love it! KAREN AI-that is brilliant. And, yes, you are right. I put the voice model through its paces in this video.

Oh no! You mean I have to stay home and play with awesome Star Wars stuff? I'm heartbroken! LOL. You're not wrong, though. What I was trying to do was imitate all of the tests done by OpenAI during the launch day. If you haven't seen the video, two of the developers ran the Sky voice through several tests. One of the tests they frequently did was interrupt the model. Thank you for the feedback. th-cam.com/video/DQacCB9tDaw/w-d-xo.html&ab_channel=OpenAI

@@DeepLearningDailyaside from that, you did a good benchmark of what it can do with GPT-4o, before the new update, since many people didn’t know how extensive this feature already has become. I think the comparison is good, to know exactly what changes in capabilities when the update does come.

@@stateportSound_wav Thank you! That is very kind. I had fun doing the video. It's a great feature, and I use it all the time. I'm looking forward to the update. They are a week late as they said it would be two weeks (for the Plus subscribers.) They are likely delayed to the "Sky" incident.

@@DeepLearningDaily yeah, another comment said they gave an update somewhere, I think blog post delaying it from the original “weeks” to “months”, but I hope that’s not the case

@@stateportSound_wav Where is this blog that says months because I never heard or read that. I think if they changed it to months, there would be more of us Plus subscribers angry and talking about that all over the internet.

The new model shouldn't need tap to interrupt. You can overlap your voice to interrupt it like in a real conversation. She didn't miss the text on screen.

Actually, they said it would be a few weeks for their Plus subscribers. I've been checking everyday. I am very excited for the new features.

​@@surfercouple they changed it to months now

@@surfercouple you have to stay updated with interviews that leadership at OpenAI do. That’s the best way to stay updated.

OpenAi clearly stated on May 13, 2024 (the day of the Livestream Keynote ChatGPT 4o launch), that ChatGPT 4o was available to "Pro", "Team" and "Enterprise" subscribers as of May 13th, but would not yet have the new "speech" functions. However, it stated that the text and image features were already working. This is not true. If anyone takes the time to go to OpenAi's webpage that showcases the examples of what this new "omni" model can do ... scroll down below the video examples to a section called "Explorations of Capabilities" where there is a drop down menu of 16 examples, showing the exact input "Prompt" and "Output" results. I tried replicating these "amazing, dazzling" results by copying & pasting the exact "Prompt" and instead of getting a mind-blowing result ... my "Pro" ChatGPT 4o output "gibberish" illegible long form handwritten text where letters were malformed, lucky if even 2 or 3 words of a 12 line poem were recognizable ... ChatGPT 4o failed miserably at replicating the example!! Go ahead try it yourself and let me know if you are able to replicate any of the 16 examples accurately.

@@satoriscope Wise advice. I will update my description to make it clear I have a Plus subscription and these features are coming to the Plus subscribers in the upcoming weeks.Thank you for the clarification. You make a valid point. I realize now that the readers of my newsletter all tend to be Plus subscribers, but that wouldn't be the case here on TH-cam since it is a much broader audience.

Me, too! I look forward to making that video!

I asked ChatGPT today and it told me it should be coming in alpha mode to plus users in the next 2 weeks. Omg how exciting, what will you try first ? Incant to test it as a tutor :)

Yes! ChatGPT is already an excellent tutor, but it will be so much better with these new features. I'd like to increase my understanding of quantum computing. But, currently, I can't interrupt the model hands free when it starts going off on a tangent with something I already know. Hands-free is important to me because I'm usually painting or sanding when I'm in my shed. (So, my hands are not "free.") This new functionality will make the model a lot more useful. I can imagine myself asking ChatGPT just about everything from now on. One note- just remember to ask ChatGPT: what was your source for that info? So, if ChatGPT told you alpha mode is coming in two weeks, then push back at the model and say: How did you know that? Where did that information come from? If the model is making it up, it will (usually) apologize. If it provides you with a source, ask for a link that you can verify yourself. I always push back at the model to make sure its not making stuff up. It is getting more and more obvious to me these days when it is, but that is probably because I use AI so much.

Hi Thomas- If it's any consolation, the "Talking Faces" technology is not out yet. Microsoft announced this technology- meaning they had created it. However, they haven't launched it. If past experience holds out, these tech companies usually wait about six months from the time they launch a white paper to the time they launch the product.

Yes, I found the pausing in the speech to be somewhat strange, especially when I tried to match up to it. I couldn't do it, even though I did seven takes of my video. But, to Microsoft's credit, this is generation one of their technology. It's only going to get better from here.

I saw this and thought: "I'm seeing the future right here. All newscasters in the future- and likely all podcasters, too- will be talking faces like this one. We will eventually regards this as normal."

Thanks for the tidbit! I'll work on an upgrade! It's probably the lighting. I was trying to match her background so I just threw up a tablecloth in the background as a mock green screen. I normally don't bother with one. The background you see behind me was just a background I downloaded from Zoom. In the real world, there is a dresser behind me in my office.

If you go to the Microsoft page for VASA-1, they demonstrate how this technology can be used with a pencil sketch or to animate the Mona Lisa. It's... unsettling. It does have a very Harry Pottery talking photos effect.

There are other technologies like this out there. Similar lip sync and head movement technology is available from AI company Runway, Nvidia’s Audio2Face AI application, Google’s Vlogger AI launched in March, and Emo AI by China’s Alibaba.

My ego thanks you!

I agree! Companies will launch a white paper, and the product will usually come out about six months later. If VASA-1 follows this pattern, we can expect this technology to be available by the end of the year. It occurred to me today that you could use this technology to bring a dead relative back to life. I'm not sure if that's a good thing or not.

@@DeepLearningDaily as I have seen in the Stable Diffusion area and using these tools... projects like this have a weird % rate of actually going from white paper to usable tool, ofc this will become a tool of sorts, but one can't be certain it will be this specific one. Knowing corpospace, it will be kept in a vault with the low quality censored weights sold via subscription services ala chatgpt and other types of platforms like that

I'm on the opposite end of the spectrum, since I'm an educator encouraging people to learn AI. It's a skill everyone can learn.

Thank you! My husband is not a fan. :) I generate the custom music for my videos using Suno.AI. There is a free version of it and you can generate song clips up to 40 seconds. Here is the "full" version of "Will AI Take My Job?" (Well, all 40 seconds of it.) app.suno.ai/song/1432b723-6965-4237-b837-dae39a11ffd2

Thanks for the feedback! I actually slowed down the speed on some of the animation. But, I'm likely guilty of moving too quickly on the still shots while trying to time the music and the images. I'll keep trying to improve on each day's video. I appreciate you providing the feedback. It's very helpful. And, you're right. There are things about this new world of AI we find ourselves in that are still very weird. Some of the art the AI generates is... odd. Sometimes I edit it and sometimes I leave it in its original form because that is what AI art looks like sometimes. It is an imperfect technology that some weird stuff, at times.