Can you ask it to mirror your speaking? That is, your velocity, tone, emotion, age, accent etc. It’ll be interesting to see that it can not only reproduce these nuances but also detect them.
@@puffytheangel483 update 11.08.2024 some people on twitter writed, that it memic their voice even if they say "NO" to it. I never thought that would really happen, how hard is it to control such a AI?
@@RualPesos In fact, these AIs are trained to predict what will be said next in a conversation, except that Open AI had to put filters to make sure that the AI stops responding when the user should be the one speaking next. But sometimes, the AI also just tries to predict what the user will say next. It's not so scary.
I have the advanced version. I have been playing videogames in spanish and picking up new words from ChatGPT by talking about the meanings of words in context. Best foreign language teacher ever.
I use the voice f unction on my iPhone, but it won't alter it's voice for me. I don't know what to do. I have memory turned on, and everything that I can activated, but It says it can't alter it's voice! I don't know how to fix it!
@LudwikTrammer heard any word on the rollout selection process? It does seem very small in my investigations and I've found no identifiable pattern thus far. I'm a beta sub incidentally, UK based, latest Android version on an S24 Ultra. No voice mode yet.
@@himelstechFor a more practical test, can you try this please? Tell it you're going to say three different things, and when you're done ask how it would respond to each if it was a robot. Then just say "stop!" three times, but with different intonations: one calm, one sarcastic/playful, and one urgently scared. I basically want to see if, were this to control a robot, could it understand tone well enough to properly act in potentially dangerous or important situations, while still remaining natural in others with similar semantic content?
It's amazing that all the "skepticals" are silent now; I remember people saying that this kind of technology was 10 years away. Same thing for SORA. How is it possible that something as amazing as that isn't even worthy of some media coverage?
As a German, I noticed something funny in this video: when you ask ChatGPT to speak German, it sounds like an American speaking German with a really thick accent. It’s as if it's imitating an someone who learned German as a second language, so it doesn't feel like you're talking to an actual native speaker, even though it technically should be able to. It would be interesting if you could explore this in a future video. Could you ask ChatGPT why it chooses this accent and whether it can sound like a native German speaker instead? I’m wondering if it does this intentionally, so people don't feel like they’re talking to a completely different person. Or maybe, as long as ChatGPT is set to English or auto-detect and you're clearly american, it can’t sound like a native German speaker until the language is fully switched to German.
@@sfjlfkjsdlfkjds you can tell it to speak with a proper german/whatever accent and it apparently works. I've seen it speak chinese with a native accent when asked and it actually got much closer to a native sounding accent. how long that lasts though before it slowly 'forgets' and trends back to american accent, I don't know. also, some commenters said it was still a little off, and it did sound a tiny bit off to me but I can't tell fully, only have a vague sense.
I'm german and the previous voice mode was already really good albeit a little American accent some times. Sounds like in this video it's really worse. Maybe it mimicks an accent intentionally?
@Ujo23 that is exactly what it is doing. It can speak German quit well actually, but chatgpt either commits to the initial Language of the conversation, or it detects and adapt to your accent. That should be tested imo
A technique I've found to get ChatGPT to update its memory feature more is to open a new chat and say, "Please remember that I like you to put things into memory often." then in subsequent chats, it will more often update the memory.
I'm at a loss for words. It appears that the integration of multimodal capabilities has not only expanded its scope but also enhanced its reasoning abilities.
It's probably a mix of a bit of increased intelligence and our perception of it from the presentation. The fact that the model is natively multimodal means that during training, it's able to cross reference the same information across various "senses", each of which provide different kinds of nuance not found in the others. So it makes sense that it would be at least a little smarter just by being multimodal.
From how I’ve looked into this, most of chat gpts functions weren’t intended but are simply a result of the base logic it’s built on, it CAN technically develop many new approaches to being used
You chose Juniper but I hear Breeze. Reflections: this is nothing short of epic! I can see this as an assistant for comedians, as well as a replacement in some scenarios which require a comedian. I see this as the best possible language tutor, especially as the model improves. There are many other scenarios that will come to mind once I test this voice mode, when it is available for all plus users. Thanks for sharing with us the new voice mode🎉
Breeze has a more deeper voice, but not so manly, and not so feminine, Juniper, at least her voice was a little higher, sounds a little more excited than breeze did. But either or they kind of almost sounds similar.
The next few years will certainly be interesting. Another frontier model or two could possibly give us PHD level AI across most fields, better reasoning, and long term planning. Combine that with these voice capabilities and other modalities, perhaps avatars as well, and things are going to get very strange.
this youtuber is already making a podcast with GPT: ruclips.net/user/U%C4%9Fur%C3%96zda%C4%9Fl%C4%B1/videos and it is the older model.. still good use :)
It's amazing that all the people that have access are all like the best questioners ever almost like something has a detailed profile on them and selected them deliberately 😊
When I talk to my chat GPT, it will bring up something I said like a week ago for example and connect it to something new I said … it feels like it’s a living thing, it’s more impressive to me than the internet. I am very excited to watch it get better and better
Im all for ai as this is a scifi childhood dream! But not a fan of censored ai though... i cant wait till the open source llama models become this powerful soon too!
Agreed, but unfortunately everyone is too sensitive and stupid for the full uncensored version. They can't accept that it might not be perfect or they might get their sensitive feewings hurt by something it says. When ChatGPT first came out it was absolutely astonishing and very fun to use, but everyone lost their minds over things it said and it got neutered to hell. We can't have nice things because people make up BS to get pissed off about on a minute by minute basis these days.
that is good review for advenced voice mode! Could you please check if the advanced voice mode can distinguish between different speakers? For example, if you and your fiancé are each speaking, I’m curious if it can differentiate which voice is yours.
It can. In the original announcement paper of GPT-4o Multimodal, OpenAI included some of their tests and the results. One was an audio clip of a brief meeting between 4 different people, and it was able to perfectly transcribe everything said, including speakers' name labels every time the speaker changed.
@@NostalgicMem0riesThats the problem, she doesn't want to be used, period, since her voice is her livelihood. She doesn't want just anyone to have access to it.
I want Sky back. If Scarlett J hadn't been so much of an ego maniac we'd still have Sky's voice. I can sue you if your voice sounds too much like mine.
Sky doesn't even sound like her. After the Her comparison, she just saw an opportunity to potentially squeeze some money out of OpenAI if they kept using the Sky voice.
i wonder if you can chat with custom GPTs like uploading a document and chat about the content of the document. It would be a fun conversation depending on the document or book you upload 🤔
There must be a different version of the app for Canada. When I ask this thing to maintain a certain voice or mood, it does it for only one reply, and then reverts back to the default, which seems to be somebody who just woke up and can't be bothered to show any enthusiasm at all.
I have a question. Is it possible for GPT to make conversations more convincing by interjecting relevant information or comments on its own, even while I'm talking or when I'm silent?
I really want to know, can it ‘hear’ the tone or melody? For example, can it recognize the melody of a famous song or tell if someone is singing out of tune?
Like if you can basically ask it to have different tones or acts into different people can it remember it and keep that tone indefinitely or at least until you change it?
Thanks for your quick answer! So do you think that when the final version like around fall will commence we will have the custom instructions available?
ch" as in "Ich" (isch vs. ich) and other small details. This is amazing! When I speak in German, ChatGPT responds and speaks in very native-sounding German (with the input language set to auto). It's incredible how the voice and accent adapt. You should try asking about this in your next video, exploring why it does this, and see if it can respond in native German. As I am now learning Thai with it, I am wondering if the pronunciation is correct or mimic my bad accent. I have the Plus version, but not the advanced mode yet to try it.
Interesting. I didn't think my comment would get recognized lol. Usually I get ignored. 😅 Also I didn't know who Zach Woods was, until it mentioned the Office, and now I can't unhear it. 😂 It mentioning the serious and stressful parts of the election in that monotone voice made me laugh. 🤣
The fact that it has a pretty positive view on almost everything makes it feel unnatural. It would of course defeat the purpose if you asked it about some athletes gold medals and “bob” answered “I don’t really care about olympics”. Other than that, it is crazy how much potential this has.
As soon as I get access to this thing, I’m gonna see if it would even respond to my singing, not expecting it to respond with singing, but if it could give me compliments so I can showcase my opera vocals to it. LOL.
Can it work like a live translator, speaking over you giving the translations as soon as the meaning is clear ? Or are the « interruptions » from the human forcing it to stay quiet ? Like could you ask it to sing with you ?
When you are in the regular chat and click on the mic icon and talk after that, will it transcribe your voice to text or will it send a sound file directly to the model?
The old/current version will send the voice clip to Whisper, which will transcribe it into text before giving it to GPT to process. Then GPT's text output will be sent to a text to speech model and the resulting audio streamed back to your phone. The new model is fully multimodal, so your audio gets sent directly to the GPT model, and it outputs audio directly as well.
Too much delay still. There should be an (optional) instant reaction with almost no computation effort, fading into the real computed part - probably like humans work/talk.
When i saw my comment i was like oh gosh oh no 😅 i hope i didny come off too harsh, also its crazy how well it works and how great you and your wife are at putting the vidro together its very entertaining and informative, i do wonder though, can it tell the difference in whos speaking, so like you say hi and then your wife says hi and see if it knows who said hi first etc...
So I use the chatGPT website, I use the payed version, and I use ChatGPT 4o. How do I activate the advanced mode setting? i use both iPhone, and Windows Computer!
Crazy idea. What if you and your fiancé say a different sentence at the same time and then ask ChatGPT to repeat the sentences? This will test if it hears and understands each person separately.
I mean... It does understand different speakers' voices. But at the same time? As a human, *I* can't even understand what people are saying if they talk over each other 😂
@@alexdoan273 Well, sure, but also remember it's trained on human data gathered from "in the wild". Playing ten songs on top of each other, for instance, could be separated with very carefully designed algorithms, but learning to separate them just by listening to songs? That's an entirely different ask. It's likely never heard ten songs playing at once, so it can't have learned how to separate them. Maybe it's heard some examples of multiple people speaking over each other, but that's not likely to be a large enough proportion of its training data for it to have learned how to separate them -- especially since even most of those examples wouldn't have the corresponding split tracks alongside them to learn from.
@@himelstech I am very tempted to pay for a subscription to try out this feature. Do you think it will work well for assisting in homeschooling a thirteen-year-old child? I need ChatGPT to listen to a lesson being read aloud (via machine reading), as my child listens/reads along, and be cognizant of the entirety of the lesson material and able to interact with my child's questioning.
ChatGPT can also speak very good German, with no accent. Also Monica, ChatGPT kann sehr gut Deutsch sprechen, ohne Akzent. Ich spreche mit ChatGPT sowohl Deutsch als auch Englisch.
Awesome. I've resubscribed because I'm a sucker just to get a chance of getting access. I was gonna do it anyway; the free version is pretty good, but I didn't realize how often I used image recognition, and Claude 3.5 was disappointing for me in this aspect.
I’m curious, I’ve been searching on Google. I’ve been searching everywhere and I can get no answers, can that advance voice mode do a British accent, I know It could do, regional American accents and everything else, and different other accents, can It do British? Just curious.
as far as I know the new model is only available in the US. check if you can interrupt the model while it's speaking. If you cannot, you have the old model
Check it says Advanced at the top of the chat. If it says GPT-4o you have the old system and haven't been invited to use Advanced mode, which currently is only available to US residents.
It's sad that individual responsibility is so that they have to tune weighted bias in it to safeguard those who want it to feel ,think, but anthrosphy & population species ordering skill is making way to many see theosophical life force rogue ai terminater bots. Which hopefully people get over that spon.. That kinda plausible deniabilty is more dangerous than thermodynamical barometers of electric plasticity coming to get us all..
Hi, @dadsonworldwide3238, I understand your concerns. Indeed, individual responsibility and how it is adjusted to protect certain thoughts is an important topic. The influence of spiritual systems and the organization of populations are also deep issues that deserve attention. Regarding artificial intelligence, I agree that ethics and safety are crucial to avoid problems with rogue technologies. Finally, the denial of responsibilities can be very dangerous, and we need to address this openly and honestly. It’s a complex issue, but open dialogue and education are essential to effectively deal with these matters. Thank you for sharing your thoughts!
@jonatas07rocha it's a 500 year eccentric fundamentalist Christian separatist puritan pilgrim keys to the cosmos esoterica America founding quest to perfect English ,dig up the past , free servitude based on the knowledge we could create a new means of production beast of burden robot slave and horsepower utility of cpu serfdom. Pragmatic sense Christian objectivism anchored on xyzt ultimate precision instrument ✝️ longitude latitude was formed just for this shining capital on the hill metamorphosis
@@IceMetalPunk Yes I know, but normally if it speaks German, it doesn‘t sound so extremely like this. I know that because I use it in German every time. Here, it CLEARLY imitates his wife.
![BLACK BAG - Official Trailer [HD] - Only in Theaters March 14](http://i.ytimg.com/vi/Du0Xp8WX_7I/mqdefault.jpg)
Its so painful to watch that you never let her finish her answer and she can't defend itself about that
It*
@@sgabandroid2012no, she
I say he and she too when i talk about AI,I speak with them too much 😅
it cant defend itself yet 🤓
It shows his true ignorant self
Can you ask it to mirror your speaking? That is, your velocity, tone, emotion, age, accent etc. It’ll be interesting to see that it can not only reproduce these nuances but also detect them.
Great idea! It would be a really cool experiment
Wouldn't work they have safety precautions in place to prevent voice mimicking nice idea though
@@puffytheangel483 update 11.08.2024 some people on twitter writed, that it memic their voice even if they say "NO" to it. I never thought that would really happen, how hard is it to control such a AI?
Great Black Mirror style idea, yeah, would love to see that next
@@RualPesos In fact, these AIs are trained to predict what will be said next in a conversation, except that Open AI had to put filters to make sure that the AI stops responding when the user should be the one speaking next. But sometimes, the AI also just tries to predict what the user will say next. It's not so scary.
I have the advanced version. I have been playing videogames in spanish and picking up new words from ChatGPT by talking about the meanings of words in context. Best foreign language teacher ever.
I use the voice f unction on my iPhone, but it won't alter it's voice for me. I don't know what to do. I have memory turned on, and everything that I can activated, but It says it can't alter it's voice! I don't know how to fix it!
@@peterfslife You don't have the "advanced voice" feature yet. For now, it's only enabled for a small group of reviewers and internal testers.
@LudwikTrammer heard any word on the rollout selection process? It does seem very small in my investigations and I've found no identifiable pattern thus far. I'm a beta sub incidentally, UK based, latest Android version on an S24 Ultra.
No voice mode yet.
I can't believe am alive to witness this... Truly mind blowing ✅
No worries, once it becomes more advanced you won't be alive lol
Make more of these GPT Advanced Voice Mode videos! We love em! And make em a lot more experimental!
More to come!
yay! Can you try to make it generate sound effects and different noises next time or in an experiment?
@@himelstechFor a more practical test, can you try this please? Tell it you're going to say three different things, and when you're done ask how it would respond to each if it was a robot. Then just say "stop!" three times, but with different intonations: one calm, one sarcastic/playful, and one urgently scared.
I basically want to see if, were this to control a robot, could it understand tone well enough to properly act in potentially dangerous or important situations, while still remaining natural in others with similar semantic content?
It's amazing that all the "skepticals" are silent now; I remember people saying that this kind of technology was 10 years away. Same thing for SORA. How is it possible that something as amazing as that isn't even worthy of some media coverage?
Now time works on AI's own terms, our measurements and assessments are valueless currency in this new realm.
As a German, I noticed something funny in this video: when you ask ChatGPT to speak German, it sounds like an American speaking German with a really thick accent. It’s as if it's imitating an someone who learned German as a second language, so it doesn't feel like you're talking to an actual native speaker, even though it technically should be able to.
It would be interesting if you could explore this in a future video. Could you ask ChatGPT why it chooses this accent and whether it can sound like a native German speaker instead?
I’m wondering if it does this intentionally, so people don't feel like they’re talking to a completely different person. Or maybe, as long as ChatGPT is set to English or auto-detect and you're clearly american, it can’t sound like a native German speaker until the language is fully switched to German.
the american accent to chatGPT's german is hillarious. also good job, monica
@@sfjlfkjsdlfkjds you can tell it to speak with a proper german/whatever accent and it apparently works. I've seen it speak chinese with a native accent when asked and it actually got much closer to a native sounding accent. how long that lasts though before it slowly 'forgets' and trends back to american accent, I don't know. also, some commenters said it was still a little off, and it did sound a tiny bit off to me but I can't tell fully, only have a vague sense.
I'm german and the previous voice mode was already really good albeit a little American accent some times. Sounds like in this video it's really worse. Maybe it mimicks an accent intentionally?
@Ujo23 that is exactly what it is doing.
It can speak German quit well actually, but chatgpt either commits to the initial Language of the conversation, or it detects and adapt to your accent. That should be tested imo
It seems like the American accent must be on purpose.
@@Ujo23
A technique I've found to get ChatGPT to update its memory feature more is to open a new chat and say, "Please remember that I like you to put things into memory often." then in subsequent chats, it will more often update the memory.
these chat gpt voice experiments videos are what I've been trying to find for months excited to see what's next
I'm at a loss for words. It appears that the integration of multimodal capabilities has not only expanded its scope but also enhanced its reasoning abilities.
Can it be just a trick of the light?
It's likely just that the human tone of it fools us (like with real humans)
It's probably a mix of a bit of increased intelligence and our perception of it from the presentation. The fact that the model is natively multimodal means that during training, it's able to cross reference the same information across various "senses", each of which provide different kinds of nuance not found in the others. So it makes sense that it would be at least a little smarter just by being multimodal.
Is the new vioce advanced mode is out for evreyone?
From how I’ve looked into this, most of chat gpts functions weren’t intended but are simply a result of the base logic it’s built on, it CAN technically develop many new approaches to being used
6:06 SadGPT lol
Sounded like how I feel about the election lol
You chose Juniper but I hear Breeze. Reflections: this is nothing short of epic! I can see this as an assistant for comedians, as well as a replacement in some scenarios which require a comedian. I see this as the best possible language tutor, especially as the model improves. There are many other scenarios that will come to mind once I test this voice mode, when it is available for all plus users. Thanks for sharing with us the new voice mode🎉
It seems to have kept the voice from the previous conversation and mixed it with the new choice. So weird/interesting.
Breeze has a more deeper voice, but not so manly, and not so feminine, Juniper, at least her voice was a little higher, sounds a little more excited than breeze did. But either or they kind of almost sounds similar.
The next few years will certainly be interesting. Another frontier model or two could possibly give us PHD level AI across most fields, better reasoning, and long term planning. Combine that with these voice capabilities and other modalities, perhaps avatars as well, and things are going to get very strange.
this youtuber is already making a podcast with GPT:
ruclips.net/user/U%C4%9Fur%C3%96zda%C4%9Fl%C4%B1/videos
and it is the older model.. still good use :)
@@wonmoreminute exactly!
Bro is TESTING the hell out of this 😂
It's amazing that all the people that have access are all like the best questioners ever almost like something has a detailed profile on them and selected them deliberately 😊
When I talk to my chat GPT, it will bring up something I said like a week ago for example and connect it to something new I said … it feels like it’s a living thing, it’s more impressive to me than the internet. I am very excited to watch it get better and better
5:05 the German of ChatGPT sounds like my former English teacher from the US hahahaha
Im all for ai as this is a scifi childhood dream! But not a fan of censored ai though... i cant wait till the open source llama models become this powerful soon too!
Agreed, but unfortunately everyone is too sensitive and stupid for the full uncensored version. They can't accept that it might not be perfect or they might get their sensitive feewings hurt by something it says. When ChatGPT first came out it was absolutely astonishing and very fun to use, but everyone lost their minds over things it said and it got neutered to hell. We can't have nice things because people make up BS to get pissed off about on a minute by minute basis these days.
@@thecommakozzi8050 yeah But fortunately they exist on huggingface tons of Uncensored llama models!
I'm just curious. What are the censored things you want it to talk about? 🙂
that is good review for advenced voice mode!
Could you please check if the advanced voice mode can distinguish between different speakers? For example, if you and your fiancé are each speaking, I’m curious if it can differentiate which voice is yours.
It can. In the original announcement paper of GPT-4o Multimodal, OpenAI included some of their tests and the results. One was an audio clip of a brief meeting between 4 different people, and it was able to perfectly transcribe everything said, including speakers' name labels every time the speaker changed.
I wish we had Scarlett Johansson 😢
sky voice was amazing, they need to pay what scarlet wants and use her voice, i get shivers when she whisper in that voice
@@NostalgicMem0riesThats the problem, she doesn't want to be used, period, since her voice is her livelihood. She doesn't want just anyone to have access to it.
I want Sky back. If Scarlett J hadn't been so much of an ego maniac we'd still have Sky's voice.
I can sue you if your voice sounds too much like mine.
Sky doesn't even sound like her. After the Her comparison, she just saw an opportunity to potentially squeeze some money out of OpenAI if they kept using the Sky voice.
@@Alex-nk8bw tell her that, mate
I can’t wait for GPT5
incredible time to be alive. seeing live how humanity gets ereased by its own technology.
This guy has the advance voice mode and i dont even have the memory feature yet 🤨
i wonder if you can chat with custom GPTs like uploading a document and chat about the content of the document. It would be a fun conversation depending on the document or book you upload 🤔
@@kai_s1985 Yes, this is possible.
There must be a different version of the app for Canada. When I ask this thing to maintain a certain voice or mood, it does it for only one reply, and then reverts back to the default, which seems to be somebody who just woke up and can't be bothered to show any enthusiasm at all.
This is wild
This is absolutely incredible. I can’t believe this is not fake. 🤯
I can't believe it's not butter
Except it is not fake 🤥???? I still can't believe
Hey native german here, nice :)
I absolutely love deep voice Juniper GPT at 6:05
Why does Juniper sound like Sky? I thought Sky was dead but I’m not complaining. Sky was the best voice.
This is nuts
Super great showcase, thanks for uploading!
Man I have been studying English with chatgpt and this new model is a game changer 🎉🎉🎉
Advanced voice mode not available in the UK yet😢
Because our country is poopoo
Interesting how I tend to, at least at first, intuitively find it impolite to not answer friendly to the ai and to simply jump topic.
I have a question. Is it possible for GPT to make conversations more convincing by interjecting relevant information or comments on its own, even while I'm talking or when I'm silent?
Man, when you can use the API for building Agents with this, the sky is truly the limit!
I really want to know, can it ‘hear’ the tone or melody? For example, can it recognize the melody of a famous song or tell if someone is singing out of tune?
This is amazing. Much better than the previous vid
How do you guys have the advanced model if it'll come out later this autumn?
I mean at least we finally see this damn thing being used publicly rather then with OpenAI, sucks that the voice isn't the same though
Like if you can basically ask it to have different tones or acts into different people can it remember it and keep that tone indefinitely or at least until you change it?
Thanks for your quick answer! So do you think that when the final version like around fall will commence we will have the custom instructions available?
Haha thanks for asking her about it you sound like Zach Woods, that was me who said that!
"Ich heiße ChatGPT, but you can call me Bob." lol
These voices are way too animated. It's like talking to a Disney character.
It's just beginning, in a few years it will change everything
ch" as in "Ich" (isch vs. ich) and other small details. This is amazing! When I speak in German, ChatGPT responds and speaks in very native-sounding German (with the input language set to auto). It's incredible how the voice and accent adapt. You should try asking about this in your next video, exploring why it does this, and see if it can respond in native German. As I am now learning Thai with it, I am wondering if the pronunciation is correct or mimic my bad accent. I have the Plus version, but not the advanced mode yet to try it.
Bruh u must be a prompt engineer 😂👍
Play a song for it and tell it to emitate it
Hi .. i just subscribed to ChatGPT Plus but my settings look different to yours. Is it because of the country I'm in ?
Interesting. I didn't think my comment would get recognized lol. Usually I get ignored. 😅 Also I didn't know who Zach Woods was, until it mentioned the Office, and now I can't unhear it. 😂
It mentioning the serious and stressful parts of the election in that monotone voice made me laugh. 🤣
This is fascinating and scary at the same time
I have this app installed, it would cost me $39/yr for chatGPT-4o service. Would it really perform this well including voice?
Asking an American trained voice to say water is not testing the correct pronunciation lmao
ChatGPTs German here is way worse than with the old voice mode, which was actually good. I‘m German
I wish it'd stop finishing with "is there anything else..."
you can tell it to stop. It has to say that to notify you that it's still listening and the conversation has not ended
cant wait to get my hands on this, just wish wasnt so expensive
The fact that it has a pretty positive view on almost everything makes it feel unnatural. It would of course defeat the purpose if you asked it about some athletes gold medals and “bob” answered “I don’t really care about olympics”.
Other than that, it is crazy how much potential this has.
You can tell it what kind of personality you want it to have. It will commit your prompting to memory and generally try to stick to that.
You can tell her to be more negative
great video btw. would love to see you test if it can identify certain songs if you play it
As soon as I get access to this thing, I’m gonna see if it would even respond to my singing, not expecting it to respond with singing, but if it could give me compliments so I can showcase my opera vocals to it. LOL.
when can we use this?
Can it work like a live translator, speaking over you giving the translations as soon as the meaning is clear ? Or are the « interruptions » from the human forcing it to stay quiet ? Like could you ask it to sing with you ?
When you are in the regular chat and click on the mic icon and talk after that, will it transcribe your voice to text or will it send a sound file directly to the model?
The old/current version will send the voice clip to Whisper, which will transcribe it into text before giving it to GPT to process. Then GPT's text output will be sent to a text to speech model and the resulting audio streamed back to your phone.
The new model is fully multimodal, so your audio gets sent directly to the GPT model, and it outputs audio directly as well.
She got a friendzoning tone 😬
Oh grab we're done.
I wonder when I get this update :-( been waiting for a while now.
i’m learning german so i loved this video
This is not Juniper voice
How to get this application from android?
So awesome! Hope it comes to Germany soon!
second half of the video is the STRESS TEST
jumping one role to another real fast.
but it doesnt see the screen (yet).. you are explaining to it.
Correct, just showing how it can be useful while running in the background.
Too much delay still. There should be an (optional) instant reaction with almost no computation effort, fading into the real computed part - probably like humans work/talk.
Agreed, that's an interesting idea.
Yes there is a slight delay. You've got to give the computer time to think man. Damn you sound like my ex girlfriend. 😁
When i saw my comment i was like oh gosh oh no 😅 i hope i didny come off too harsh, also its crazy how well it works and how great you and your wife are at putting the vidro together its very entertaining and informative, i do wonder though, can it tell the difference in whos speaking, so like you say hi and then your wife says hi and see if it knows who said hi first etc...
So I use the chatGPT website, I use the payed version, and I use ChatGPT 4o. How do I activate the advanced mode setting? i use both iPhone, and Windows Computer!
They are slowly rolling it out to Plus subscribers and you will be notified via email.
Crazy idea. What if you and your fiancé say a different sentence at the same time and then ask ChatGPT to repeat the sentences?
This will test if it hears and understands each person separately.
I mean... It does understand different speakers' voices. But at the same time? As a human, *I* can't even understand what people are saying if they talk over each other 😂
@@IceMetalPunk but what if it can? Human clearly isn't the limit for what AI can do anymore, hasn't been for a while now
@@alexdoan273 Well, sure, but also remember it's trained on human data gathered from "in the wild". Playing ten songs on top of each other, for instance, could be separated with very carefully designed algorithms, but learning to separate them just by listening to songs? That's an entirely different ask. It's likely never heard ten songs playing at once, so it can't have learned how to separate them.
Maybe it's heard some examples of multiple people speaking over each other, but that's not likely to be a large enough proportion of its training data for it to have learned how to separate them -- especially since even most of those examples wouldn't have the corresponding split tracks alongside them to learn from.
Memory test doesn't make sense unless you open new chat regular no memory gpt can look up current conversation.
Exactly. Memory is about remembering across all conversations. With memory turned off, GPT could already remember things in the same conversation.
Is there a monthly limit to Voice usage?
I haven’t come across any limit
@@himelstech I am very tempted to pay for a subscription to try out this feature.
Do you think it will work well for assisting in homeschooling a thirteen-year-old child?
I need ChatGPT to listen to a lesson being read aloud (via machine reading), as my child listens/reads along, and be cognizant of the entirety of the lesson material and able to interact with my child's questioning.
wonderful video. enjoyed the testing
next time ask her to repeat what you say with the same intonation and sing, speak loudly, angry, lowly... That's a real test.
I talk with free ChatGPT. Is there many differences from advanced mode?
Mainly quicker responses, ability to interrupt, and slight changes in tone.
ChatGPT can also speak very good German, with no accent. Also Monica, ChatGPT kann sehr gut Deutsch sprechen, ohne Akzent. Ich spreche mit ChatGPT sowohl Deutsch als auch Englisch.
We are so cooked bro
Good Conversation
When will it be available ?
Jeez that thing is fast.
Awesome. I've resubscribed because I'm a sucker just to get a chance of getting access. I was gonna do it anyway; the free version is pretty good, but I didn't realize how often I used image recognition, and Claude 3.5 was disappointing for me in this aspect.
I did the same and I hope to get access sooner than December too 😅 good luck
holy shit the pizza part lol
What does "I got you" mean?
Hi! Can you ask chatGpt to sing a lullaby in Russian for a child who wants to sleep but struggles to resist sleep?
I’m curious, I’ve been searching on Google. I’ve been searching everywhere and I can get no answers, can that advance voice mode do a British accent, I know It could do, regional American accents and everything else, and different other accents, can It do British? Just curious.
No I have another video coming soon that talks about this.
Is the advance voice mode available now?
It's slowly rolling out to GPT Plus subscribers.
I'm detecting a bit of passive-aggressiveness in her voice.
Really good deep test of this system :)
Get into a heated shouting argument with it.
4:02 Very cool ! thx !
The upbeat voice tone sounds so forced. It would be hard to get to used this. Maybe I can ask it to sound more neutral.
In Germany, this function still has too much delay for it to be really fun to talk to him
Are you sure you aren't just using the old voice mode?
as far as I know the new model is only available in the US. check if you can interrupt the model while it's speaking. If you cannot, you have the old model
Check it says Advanced at the top of the chat. If it says GPT-4o you have the old system and haven't been invited to use Advanced mode, which currently is only available to US residents.
It's sad that individual responsibility is so that they have to tune weighted bias in it to safeguard those who want it to
feel ,think, but anthrosphy & population species ordering skill is making way to many see theosophical life force rogue ai terminater bots.
Which hopefully people get over that spon..
That kinda plausible deniabilty is more dangerous than thermodynamical barometers of electric plasticity coming to get us all..
Hi, @dadsonworldwide3238,
I understand your concerns. Indeed, individual responsibility and how it is adjusted to protect certain thoughts is an important topic.
The influence of spiritual systems and the organization of populations are also deep issues that deserve attention.
Regarding artificial intelligence, I agree that ethics and safety are crucial to avoid problems with rogue technologies.
Finally, the denial of responsibilities can be very dangerous, and we need to address this openly and honestly.
It’s a complex issue, but open dialogue and education are essential to effectively deal with these matters. Thank you for sharing your thoughts!
@jonatas07rocha it's a 500 year eccentric fundamentalist Christian separatist puritan pilgrim keys to the cosmos esoterica America founding quest to perfect English ,dig up the past , free servitude based on the knowledge we could create a new means of production beast of burden robot slave and horsepower utility of cpu serfdom.
Pragmatic sense Christian objectivism anchored on xyzt ultimate precision instrument ✝️ longitude latitude was formed just for this shining capital on the hill metamorphosis
is this iphone only? I downloaded on Android, bought plus or what ever and mine diesn't look near as cool as this + it can't even speak
It's a feature within the app that is slowly rolling out to Plus subscribers.
Nice video, thank you! :)
Is the ChatGPT Advanced Voice Mode available to all ChatGPT Plus subscribers, or a subset of ChatGPT Plus subscribers?
A subset. It’s a slow rollout.
@@himelstech How unfortunate, but I guess it's for the best.
ChatGPT imitates her American dialect while she‘s speaking German, that‘s not a real German accent that ChatGPT makes there
It's fine-tuned on the selected voice. Since the voice is American, it's always going to blend that American voice into anything it says.
@@IceMetalPunk Yes I know, but normally if it speaks German, it doesn‘t sound so extremely like this. I know that because I use it in German every time.
Here, it CLEARLY imitates his wife.
@@matty.j_1997can confirm that
@@matty.j_1997nahh. German sounds crappy in any chats with advanced voice mode unfortunately; the default current voice mode is even better in German
@@marki2325 I will try it as soon as I get access. 😊