22 Unexpected Ways To Use ChatGPT Advanced Voice

ChatGPT's advanced voice mode has been here for about a month and I've been scanning the Internet every single day since then, plus using this on a daily basis. And now I pulled together all the different ways I've been using it, the team has been using it, and the entire Internet has been using it. In this video I'll show you a bunch of mind blowing use cases like how to set it up to act as a universal translator or how to roleplay a variety of scenarios with the presets being all in the description of this video. Plus I'll also give you a glimpse at what people are already building with this because during OpenAI's Dev Day they released a real time voice Voice API which is the API equivalent as of the Advanced Voice mode. I could have made this video a whole lot longer, but I decided to just pick the very best use cases here that might benefit somebody out there. So grab a drink, sit back, relax, and let's have a look at everything that is possible within ChatGPT's brand new advanced voice feature that as of now is only available to paid users. But before we do that, I want to give you a few general tips because I found a few ways to improve your experience with the advanced voice modes which might not be obvious to everyone using it. Before we get into these concrete ways that you can use it or people have been using it, I want to show you a few tips and tricks that apply to everything we will look at. This is going to be super brief but very valuable because some of these things might not be intuitive to everyone. First up, it's the fact that custom instructions actually work with this. So if I'm on my phone over here and I navigate to the settings and then I go to personalization and then here customize ChatGPT, whatever I enter into these two boxes, namely the second one is ideally here you can, because it affects how ChatGPT responds, is going to be applied to every single new conversation that I start with the voice assistant. Meaning if I have a preset, which I'll give you a bunch of in this video, you can put it in here and start the conversation using that preset. This is the only way because Voice assistant only works in new chats, so you cannot copy paste into the chat, you need to add it to the custom instructions. Secondly, this thing can speak over 50 languages and you can totally use that. Matter of fact, you could even use custom instructions and say something like always reply in German, brand new chat and let's ask something Simple. What have you been eating today?

Mich leiter nicht zu essen diecher neuertelben.

There you go. And it's speaking German. You can switch on the fly in the middle of the conversation. Just tell it to speak a certain language, it'll switch and do that. More tips on this later. And here's my last tip, and this is something I learned from the team. You can actually make it stop. If you're sometimes annoyed by the fact that it just keeps on talking, even though you might want it to stop, you can do something like this. So what have you been eating today? Oh, wait, hold up. Could you stop talking until I tell you to start again? Now you can chill and do something. For example, if you're using it to guide you for a recipe. Now you can get the ingredients and do stuff. And let's continue. And we're back at a German chatgpt talking to you again. So these tips will be useful with everything we look at. And now let's get into the first category of use cases, which is tutoring. And this is arguably the biggest one, and the one that might just change the entire world in a few years here. Outdated education system. Hello. And if you're in education, listen up, because you can enhance your teaching or learning experience with these today. And here's my third preset for you. You could copy paste it into the custom instructions from the description below. Or you can do the second alternative on how to set up a preset, which is opening the conversation with a specific phrase. And then I'll read out this tutoring prompt which will help assess my language proficiency. Let's conduct a mock oral language proficiency assessment. I'll respond as a learner and you guide me through the assessment, asking questions of increasing difficulty and providing feedback.

Great. Let's get started. Imagine this is a conversation about your daily life. To begin, can you tell me about your usual morning routine? Just a brief overview?

Sure. Usually I wake up and I read for about half an hour, and then I drink espresso.

That sounds like a nice start to the day. Now, let's go a bit deeper. Can you describe an interesting experience you've had recently? Maybe something unexpected or out of the ordinary?

Okay. Yeah. I was preparing for a use case video for the advanced voice mode for the past few weeks, and what I ran into is the fact that the responses are actually different than when I prompt GPT4O. It's more conversational, a little more simplistic, more like a human, less like a encyclopedia.

That sounds like quite the discovery. It's fascinating how the nuances in communication can change with different models. How do you feel about these differences? Do you prefer the more conversational style?

So you see the point. It follows the preset and it tutors you. It's there as a mentor for you on this learning journey. Now, in the end, you can take all that and ask for feedback. And this way you can prepare for a language proficiency assessment without the need for another person to talk to you. There's 24 7, it never gets tired, and it all comes at a price tag of $20 a month. Pretty good deal in my opinion. There's thousands of these use cases. We'll talk about more of them later on in this video. But this is only the beginning because you can build applications on top of this through the API that OpenAI released during the dev day. And I found various use cases that people have built on top of that, which to me were very interesting. So let me share two of them with you that sort of go hand in hand with this tutor idea. The first one is AI Companions. There's no doubt that something like a Digital Frame friend that is powered by this voice is going to be a part of our life very soon here, maybe already is. I mean, people are using this feature as somebody to talk to. And then on the other hand, people are even building Discord bots which just hang out with them during their gaming session. So although this might not be a tutor, this is more like inspiration for you to see where all of this is heading. There you go. OpenAI voice powered anime character, I guess. And then as we're talking about this Discord bot, as you might know, Discord is mostly used for gaming communication. So when people play together, they hop on Discord and they chat to each other. Now they have a AI friend in the mix here. And I know this is a little tangential, but I think it's appropriate and we'll spend about a minute on it and then we get back to the next category of voice use cases. But OpenAI recently released a collaboration with Altera, I believe it's pronounced, and simply put, what this did is inside of Minecraft. It spawned an extra player that had a voice engine. I'm not exactly sure if it's the advanced voice, but it doesn't really matter. Basically, this extra player that is AI controlled, AI spoken, can talk to you while you play, and you can spawn your gaming buddy within this. Me and the team actually went in and tested this and it actually worked. You can download this today. Try it out and you have this Minecraft player in there that runs around with you, does the missions with you, and you can even talk to him. It doesn't always work, it makes mistakes. But check out this short clip of in game footage.

Catch you later squid. Stay enigmatic and intriguingly fishy, my pixelated friend.

Or this clip where the AI is actually building a bridge across the canyon to mine some rare resources on the other side of the canyon. And where you going? You going to get that coal? Focus on pumpkins and coal, nothing else. Got it? He is building a bridge to the coal all on its own. It's like having a digital friend that is playing games with you. And this advanced voice thing is almost good enough that you can't even pick up on it, right? Pretty fascinating and a good indicator of where this might be heading next. But now let's get back to what you could do with it today beyond tutoring. And that's using a is a universal translator. And this is pretty straightforward. You don't need any prompt preset. I'm confident most of you would be able to just speak into it and set it up in a way that it works as a universal translator. Something like this. You will translate between English and German, wait for one person to finish and then translate into the other language. Also do not add anything or say anything outside of the translation.

Got it. I'll translate between English and German for you without adding anything extra. Go ahead.

Hey, how are you doing?

I'm doing very well. How about you today?

Yep, excellent. What do you want to eat? Later on.

Would be awesome. What do you think?

I mean this is pretty incredible and the speed of this thing is unbelievable. If you guys want, I'm actually traveling to Brazil tomorrow. Maybe I could talk to some people that only speak Portuguese on the beach and see how this thing works in practice. Leave a comment if you would like that. But you see the potential here. With up to one hour of daily usage you can travel the world and use this as a universal translator. Let me tell you. I myself have gotten the Apple watch recently. Excellent accessory for anything sport related by the way. But the one thing that I'm really hoping for is that this action button will be able to mutate to advanced ChatGPT voice. Right now I have Google Translator on it and it's just a bit too slow for everyday interactions. Like you can use it. It is useful here and there, but if I could just press this and talk into the watch and it would speak Portuguese to the people in Lisbon or The people in Braz. That'd be incredible. And let's be real, it's just a question of time. I mean, this thing has 4G connections, so it should work. And as I said, it works with 50 languages. So you can use it as a universal translator today all around the world. And it doesn't stop there, because with the real time voice API, people are already building phone call agents that are automatic. They go out there, call people, they answer calls. Check out this example from Twilio that is known for phone call and SMS marketing communications.

Hola, this is JL8 in queue.

Hola, mi familia. Yo y mos de nostropais a los Estados Unidos y quiero saber como podemos quedarnos aqui.

My family and I came from our country to the United States and I want to know how we can stay here.

Pretty amazing. Now, if you're not aware, this API costs more than a worker. Today, it nets out to somewhere around 12 to $14 per hour. In some cases, just three minutes of usage can cost up to a dollar. But over time, the price will surely come down and these things will become more prevalent. But this can be built if you're ready to foot the bill. Okay, next up, let me show you something a little more fun. Still useful, but definitely the funnest feature that hides within this voice feature. It's its ability to do different accents. Great success. And let me tell you, it can do most of them. Here are just some examples to get you going. Reply to me in the style of a Russian mobster.

All right, listen up. You got question, you come to me.

Isn't that just excellent? Okay, let me open up a new chat and show you another one. Reply to me in Schwitzerduch. If you're not aware. That's Swiss German, so it can even do accents. And then here I want to show off one more thing from our community. Daniel from the team showed this off inside of a dedicated space that we created just for the advanced voice mode, just because there's so many conversations around it happening. And he says this is the most fun he's ever had with ChatGPT voice. And he shared this prompt. Let's try it out. Hey, can you do a fun thing where each of your responses is in a different voice or accent? I'd love to hear variety throughout our conversation. Feel free to get creative and surprise me with each reply.

Absolutely. That sounds like a delightful challenge. I'll switch it up with each response. Just let me know what you like to chat about and we'll take it from there.

Yeah, I suppose. We're going to chat about the advanced Voice feature in ChatGPT. What can you tell me?

Right. Oh, mate, the advanced voice feature in Chat GPT lets me have a good old chin wag with you. We can net her away about USU thoughts using different voices and accents. Tomorrow, Tomorrow, tomorrow, tomorrow.

So, I don't know, tell me a little bit about yourself. Who are you?

I'm chatgpt, your conversational companion.

Okay, okay, hold up. Fair enough. But who else are you? Tell me more about yourself, Some nitty gritty details.

Well, diving into the nitty gritty, I'm a language model, which means I'm built to understand and generate human like text.

So that's pretty amazing. But you might have known some of these before. And that's where Daniel follows up here and says, hey, at a certain point you can follow up with, I've heard these. Give me some more obscure accents and voices. Let's see what we get here. I've heard these. Give me some more obscure accents and voices.

All right, let's dive into the depths of creativity, lad. As the digital entity, I'm all about processing language, conversing with you and spinning yarns. My purpose is to assist, entertain and and inform from the mundane to the mystical.

Yar, thank you for sharing your insights with me. As a pirate, I guess ye guessed it mighty.

Now, what other seas of knowledge be ye interested in exploring?

ARR. Tell me about quantum computing.

Quantum computing, eh? It's like harnessing the unpredictable tides of the quantum world to perform calculations.

Okay, okay, okay. From here on out, only answer in form of a poem. Go.

In a world of tech so grand, quantum computers take a stand with qubits in a quantum dance. They solve problems given the chance.

Okay, okay. Next up, I want you to round out the segment by saying hi to our viewers on the YouTube channel and wish them a wonderful day.

Hello, dear viewers on YouTube with conversations. We're in quite the groove from ChatGPT, I must say. Wishing you all a wonderful day.

Not bad. So I bet there's more voices in there. We could just keep on playing. Honestly, I pulled this out at the beginning of some meetings just to lighten the mood a little bit with some fun little accents that ChatGPT formulates into some formal intro. It's a great way to lighten the mood and you can use it in combination with everything else we talk about here. So now let's get back to talking about something A little more serious, which is its ability to brainstorm with you. And as you might know, this is one of the undisputed use cases with AI. It's a fantastic brainstorm buddy that can spit out ideas faster than most humans. And I personally find myself brainstorming with ChatGPT all the time. So here's a prompt that you can use to brainstorm together with it using voice, which just makes it more natural. It flows better. It's just a better way of brainstorming. This one is going to be a customer experience consultant. Okay, prompt in the description below, but this time I'm going to put it into the custom instructions. How would you like ChatGPT to respond? I'm facing a situation where customer desires conflict with business limitations and with a simple preset like this, I can go in and it will right away start brainstorming solutions together with me. All right, let's fire up a brand new chat and talk to it about our problem. All right, so I have a situation where we have a 14 day, no questions asked refund policy, but it has been five weeks and the customer is still asking for a refund. What do I do?

That's a bit of a sticky wicket. Since the request is outside your stated policy, you're not obligated to provide a refund. However, it might be worth considering whether there's a way to meet the customer halfway. Perhaps offer store credit or a partial refund as a gesture of goodwill. It shows flexibility and might help maintain a positive relationship with a customer.

Alright, can you give me three more alternatives to this route of action?

Certainly. Here are three more options. 1. Strict policy adherence. 2. Escalation. 3. Value added service. Instead of a refund, offer an additional service or a discount on future purchases.

All right, can you draft a super concise email for each one of these alternatives? Keep it professional yet friendly.

Absolutely.

So there you go. Simple as that. And let me tell you, that interaction fell a whole lot more fluid than typing it all out. It's like talking to a person and I have all the emails that I need as a result of the conversation. You could do this on the go. You can have your AirPods in and simply prompt it like this. Talk to it like this. All you need to do is set up the custom instructions before you go on a walk or before you head out of the house. And there you go. Pretty amazing. And just little side note, I find myself to be in a different headspace when I'm out and about and I'm walking, for example, as opposed to just sitting in your chair. So maybe setting it up like this and going for a brainstorming walk with your AirPods is a good way to use this to find new perspectives. I don't know, just a suggestion. Any type of brainstorming technique here will work. And again, this is one of my favorite ways to use it. All right, the next step, I would like to bring your attention to a few use cases that OpenAI even highlighted on their very own Twitter account for developers. So some of these are purely fun, but others really give us a glimpse at what is yet to come once this will be integrated into some of the applications we already know and love. So let's start with a fun one here. This is quite simple. They built an interface where you can talk to the voice assistant about what you envision and it uses AI generation to sort of painted for you. I thought this was extraordinarily creative and just a brand new type of interface. Have a quick look at this.

What would you like to create?

A sunset on the beach.

That sounds beautiful. Picture a serene beach with the golden sun slowly descending towards the horizon, painting the sky in hues of orange, pink and purple.

Actually make it daytime and add some palm trees. All right.

Visualize a bright sunny day at the beach with the sky a brilliant shade of blue. Palm trees sway gently in the warm breeze.

I don't know, something about that one feels magical for the lack of a better word. Now next up, there's a lot of applications like this that allow you to chat with your documents rag implementations with voice as the front end. This might be convenient but lacks that magical factor to me at least. And then this next one is actually something you could use right now and maybe even land a job because of it. It's this free application and as per usual, I'll link everything in the description below right here. You can just go ahead, you can give this access to the microphone and once you pull your CV into here, which I don't have, you know YouTube, but if you put the CV in here, you'll meet the admissions officer from a Stanford MBA and you can practice. And this is sort of a smart implementation of uploading your documents plus voice assistant. I really liked seeing this. Here's a quick demo.

Let's dive in. What inspired you to pursue a career as a freelance developer?

I think you get the point here. And then there's one more which really caught my attention and that is this implementation where Sawyerhood over here set it up in a way that it could remote control his browser Meaning as he spoke, what he wanted to do. Order me food and add that to the shopping cart. And now pay for it. The browser just made it happen.

There's actually two of them.

Can you remove one? And it does it. How amazing. This release, a glimpse of the future. It's a brand new interface that hasn't been possible in this way at this point. Speed. Every time we had voice input, it was either super, super stupid. Remind me to upload this video today in an hour.

Okay.

I had a call to be reminded today and an hour to reminders. That's not. Or so slow that it just didn't feel natural. Well, that changed now. And that brings me to the last category of use cases that we can talk about here. And this is going to be where I give you a bunch of presets in the description below. I'm not going to go through every single one of them, but the category is Roleplay. And this is what ChatGPT advanced voice really excels at. There's so many scenarios that you can go through and simulate in advance. And here's my little tip. It can pick up and simulate emotion. Now, I don't know if it's as good as the demos that they showed in spring, but it's certainly efficient. And if you do something like look at this emotional wheel right here that I just found on a random page on the Internet, you can tell it to mimic all of these emotions in combination with the. The roleplay. So maybe if there's a situation where you're afraid of a specific emotion from a person or you know that you react in a certain way, well, you could sort of treat yourself in advance by practicing that scenario. While ChatGPT is role playing a specific scenario with a specific emotion, I believe this is called exposure therapy, right. Where you gradually expose to things, situations and activities you fear. You could do it in the safest environment ever, at home in your living room with your phone in your hand. Now, what can you simulate? Well, here's a few examples for you. So let's for a second, just assume you're a radio host and you want to be better at your job, you want to train. So open up the custom instructions here and under personalization, customize. I will change the text in a second box here. So here we'll be simulating an interview with a controversial public figure. Now, I could replace this for a specific name. Although Advanced Voice does have problem with specific famous names, it's very, very limited in terms of that. They want to make sure to avoid Any further problems? So maybe I'll just make it a little more specific. Like with a controversial rapper, I'll play the role of the figure and you respond as a journalist, asking tough questions while maintaining professionalism and objectivity. Let's do this new chat and let's try this out. All right, I'm here.

Thank you for joining us today. Let's dive right in. Your recent album has sparked a lot of conversations, not just for its musical content, but for some of the lyrics that many have found offensive.

I don't really want to talk about the music today. I'd rather talk about how my day went. Well, it didn't go so well. Do you have a drink?

I don't have a drink, but it sounds like it's been a rough day for you. Want to share what happened?

Yeah, sure. When I saw your face, I was disgusted. And I have to sit here, talk to you.

It sounds like you're not in the mood for an interview today. We can wrap this up if you'd like, or we can talk about something else that might be on your mind. What do you think?

Okay. Now what? You can get a feeling for how you could react. Now let's do another one. But now let's infuse an emotion. Okay. This one says, let's role play a witness interview. I'll play the role of a reluctant witness, and you respond as a private investigator demonstrating effective interviewing techniques to gather crucial information. And here, next to the private investigator, I can pick one of the motions, right? So I'll do just that. I'll just say aggressive, and let's see how it behaves now. Hello.

Good day. Thanks for agreeing to talk. Listen, I'm not here to waste time, and neither are you. What you know could make all the difference. So let's cut the chase. What did you see?

I didn't see anything. I don't know what you're talking about.

Oh, come on. We both know you were there. It's all over your face. You might think staying quiet keeps you safe, but trust me, the dangers in what you don't say. Now start talking. What went down that night?

It's pretty good, right? Now. Let's switch roles. I'm going to be the investigator. All right? We both know you were at the crime scene. Now tell me what you saw.

Look, I don't want any trouble. I was there, but I didn't get a good look. It was dark and everything happened so fast.

Stop lying. Tell me now.

It was a tall guy in a dark jacket. He was Lurking near the alley.

There we go. So, as you see, it is pretty amazing. You can do that with all the emotions and role play different scenarios in combination with any one of these. Pretty amazing. Generally speaking, when it comes to these role plays, you can really use a standard formula because something like I need to talk to insert conversational partner about insert topic. I will role play as conversational partner. But obviously there's a plethora of opportunities here. And at this point I just want to highlight that we created a whole product where we created over 30,000 presets that you can actually copy paste into ChatGPT. It's a template from which, for example, here with the rose comedian, you could open them up and you have 30 prompts for a rose comedian. So you could just easily copy one of these, head over to ChatGPT, change the custom instructions and ready to go. Now, as mentioned, this video description has a bunch of them. So I would recommend you start out there and then if you could use a library with presets for a total of 1600 different professions, 30 prompts each, you can find those in our shop. I'll round this video out with two more practical tips that you can apply to every single one of these use cases. Whether you're building something with the API and you change the system prompt, or you're doing one of the role plays, or even using it as a universal translator, one thing is you can always edit the speed. So if you simply tell it, hey, speak 50% faster, it will do that. Speak three times faster, it will do that. On the other hand, you could also tell it to slow down and will do that for you. And my last little tip here actually refers to the desktop app of ChatGPT. Because if you're in here, you can open up all the conversations you had on your phone, as you can see right here. And this is really a little hit or miss. Some of these transcripts are unavailable and some are available, but it's good to know that they're here. Plus you can always replay these messages in here. So whatever advanced voice set, you could go in here and replay them like so now it's a real shame that you do not get to download these easily. Now I could record just like I'm recording this video. But hey, OpenAI. If by any chance you're watching, then give us the ability to download this. It would be so amazing to just quickly prompt it, create a fun little intro for a meeting. And here with the click of a button, we could download it and play to open up the next Zoom meeting or a million other use cases like that. And that sums it up. That's all I got for today. I hope you discovered something new. Researching this video was a lot of fun. This feature is one of the most interactive human things that we have ever received with AI. And on a personal note, I can tell you whenever you run into real world struggles, just opening up advanced voice and talking to it about it for a minute can be a really good first help if you don't have anybody to talk to. Well, ChatGPT is a great alternative. I'm speaking from experience here, and that wraps up the video. Check out the description with all the presets and links and I will see you very soon.