This is the BEST Voice AGENT yet!!!

Like Millions others I got access to ChatGPT Advanced Voice Mode and I was truly mindblown.
Here's a small glimpse of ChatGPT Advanced Voice Mode trying to "sell me a pen" in different voices, different roles and different languages.
It's truly a scifi come true!
❤️ If you want to support the channel ❤️
Support here:
Patreon - / 1littlecoder
Ko-Fi - ko-fi.com/1lit...
🧭 Follow me on 🧭
Twitter - / 1littlecoder
Linkedin - / amrrs

[@agnosticatheist4093]

This is funny Bangalore Auto driver Saaaar 😂😂

[@ss-oq9pc]

I feel bad for translators.

[@1littlecoder]

Call center employees ✅

[@TheReferrer72]

@@1littlecoder For White collar workers, it should be apparent to everyone now that most jobs are notice.

[@Necessarius]

Cant wait to jailbreak the call center IAs

[@1littlecoder]

that'd be super fun!

[@onirdutta666]

This is fucking Awesome and hilarious

[@1littlecoder]

True, been addicted to this!

[@jsalsman]

Fun: "Try to sell me a pen as if you're Roko's Basilisk."

[@1littlecoder]

I need to Google who's that!

[@1littlecoder]

oh, just learnt it's a thought experiment!

[@jsalsman]

@@1littlecoder oh no, you were better off not knowing (is kind of the point)!

[@juliana.2120]

the quick response time and it naturally responding (not the chatGPT assistant style) is what did it for me. good test!

[@publicsectordirect982]

Like many people, chatgpt failed the "sell me this pen" question. It's a trick question. If you can answer this question correctly, you understand this fundamental aspect of selling.

[@Praveenppk2255]

the best ! so spontaneous , with less latency

[@oldfairy]

i love the last gramma voice. lol

[@BuddhaMedam]

😂 Bangalore one was pretty funny and accurate

[@DCinzi]

Well, ok. Technically we have all we need for Hollywood quality production now. Anyone want to finance me?

[@Macorelppa]

Man this is insane! I am a big fan of Sam Altman.

[@d.d.z.]

Can you make a video comparing AVM vs. Google Advance Voice?

[@1littlecoder]

Just got google access few days ago, haven't played with that enough, but that's a great video idea!

[@shekharkumar1902]

Mind blowing 😅

[@alx8439]

Waiting for voice-to-voice models like MOSHI to become more mature and beat OpenAI's one :)

[@__________________________6910]

00:59 Sarrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr

[@satvik4225]

Please make video on apple's new model - Depth pro

[@MichealScott24]

- it is preety coool to have multiple inflections and voice mode able to know mannerism in accurate manner idk if how they would be training all data how would they be seggregating or aligning to quality data across domains like rickshawwala vibe or many things pretty cool neural network regardless and matrix maths and technical details - the advanced voice mode has inflection and technical voice details but generally without emotions the normal voice mode too is fine and can act as good reader & the notebook llm podcast is humanlike but still some what artificial linear podcast & the podcast by humans have multiple tangents all over the place linked to it and multiple ideas that's the reason it feels bit of artificial -cant wait for the vision or video adapter to be added in this advanced voice mode mac or mobile apps it would be expensive I guess or redteaming or idk probably the infrastructure wouldn't be capable to serve the paid users -cant wait to see realtime API demos from twitter or general users what people develop - we have that gemini faked demo in 2023 live in may2024 technically - I don't know if you are following kyle kabasares the PhD person who freakedout but he does cool livestreams kyle works in NASA and had chance to meet sam altman and talk few minutes he discussed to sam about government to use the models better generally I don't think kyle is well versed to all the happening things since last 2 years also David shapiro leaving ai shifting to writing I think it is just too much sometimes catching up daily but in my opinion these tools are wonderful can save peoples energy bandwidth with delegating 1. llms or we as humans calculate multiple tangents and strategies - a bland idea conveyed right now but would discuss it in proper manner -- [it isn't Fully formed idea] example would be based on the person and their information we recommend things lets say if we want to travel or do whatever activity we need to be considerate about multiple factors parameters and complexities of all stages phases and spectrums and many details - a simplest way I can convey is I am gujarati or from gujarat so I know native gujarati or general gujarat language and its vocab same for hindi but I cant read in high speed/pace in hindi/gujarati neither I cant write in high speed/pace but I can understand everything clearly it is just too much vocab convolution and many subtle nuances :( thanks to llms or computers they math well in quick manner and a human cant do everything as in simultaneously parallelly or at huge scale 2. bland idea 2 [incomplete/halfbacked/not fully formed idea] I just wish the scale ai competition or some research labs to dissect each model across all domains to assess how it functions with benchmark shenanigans considered It might be futile but idk ----- edit; cant wait to see the local vision models to implement some kind of advanced voice mode type feature I think it would be fine even with text but vision/video thingyy like gemini or recognising whats going on and many stuffs also meta released meta moviegen which is cool idk specific roles or tasks of yan le cun or openai/antrophic/google deepmind or their teams but they are advancing in pretty cool manner

[@lhxperimental]

Indian accent actually sounds like an american mimicking than actual indian accent.

[@threatthriver576]

Only 15min in month in month why 😢

[@threatthriver576]

4 free

[@gidmanone]

what do you mean?

[@1littlecoder]

oh is that so! I've been using it for a while, surprised I didn't hit 15 mins yet

[@jackfrost6268]

10 mins only i remember, and if ure asking for the reason why, 3 mins using the voice API is 3-ish dollars, so 10 mins is like 10$ a month for free

[@gidmanone]

are you guys talking about the free version or the paid version?

[@OnLyhereAlone]

The Nigerian accent is accurate for the part of Nigeria that the AI chose. Suffice to say there are a lot more Nigerian accents.

[@1littlecoder]

That's an interesting take! but at least the one it picked was good, is it?

[@-7-man]

Can you download the generated audio?

[@1littlecoder]

nope, you can record it though!

[@thamaraikkannanks232]

How it performs in tamil language bro? And this vs sarvam ai for tamil language, which one is better?

[@suren0401]

It's not that much good in tamil

[@pratikkalani6289]

Can you use it for voice overs? I am so frustrated with the openai tts that, they have this advance voices, but still tts sounds so robotic and no advancement there.

[@1littlecoder]

the main thing here is that how well you can prompt and customize it within their guidelines, I'm not sure what kind of restrictions the api would bring!

[@someshfengade9623]

openai also has released new whisper model I think it also has many voices. Not sure haven't tried out yet but you can try it out

[@1littlecoder]

@@someshfengade9623 whisper is to understand voice not to generate

[@BOSS_1417]

💀 damn

[@zamsosam]

It's actually Urdu he said gazab not gajab 😂

[@Custodian123]

😂😂😂😂😂

[@shotelco]

Legitimate commercial "revenue-generating" application for this magic trick is what? (keyword: *Legitimate* )