The Voice AI Nobody Expected (AI News You Can Use)

Подписаться 262 тыс.

Просмотров 30 тыс.

50% 1

A new AI called Moshi just came out of nowhere and claims to do real-time voice conversations like GPT-4o's voice mode, but how good is it really?
Plus Runway Gen-3 Alpha is out, I spent a lot of time testing that and you need to see what it's capable of. All that and way more in today's video.
---------------------------------------------------------------------------------
Links:
us.moshi.chat/?queue_id=talkt...
---------------------------------------------------------------------------------
Prompts:
00:00 What’s New
00:23 Moshi AI
05:06 Runway Gen-3
11:34 ElevenLabs Reader App
12:43 Suno AI Mobile App
13:05 Luma Keyframes
14:34 Motorola Add
15:36 Pro Search
16:03 AI Fun Stuff
17:40 Dolphin Vision 72b
18:34 Figma AI Updates
20:13 Google Crossword
20:59 Open LLM Leaderboard
#ai #ainewsyoucanuse
---------------------------------------------------------------------------------
🔑 Get My Free ChatGPT Templates: myaiadvantage.com/newsletter
🎓 Join the AI Advantage Course + Community: myaiadvantage.com/community
🤯 Unlock ChatGPT's true potential: shop.myaiadvantage.com/produc...
🐦 Twitter: / theaiadvantage
📸 Instagram: / ai.advantage
🛒 AI Advantage Shop: shop.myaiadvantage.com/

Наука

Опубликовано:

15 июл 2024

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист

Посмотреть позже

Комментарии : 107

@dijifakeye2522 10 дней назад

“Im just a machine”🎀🥺

@joannot6706 10 дней назад

Worse, "I'm a just machine"

@tekmepikcha6830 10 дней назад

That Moshi interaction was funny

@EyeWillPlay 8 дней назад

Lmao. It’s like an ex-wife

@Bundles06 7 дней назад

Didn't expect this video to be the funniest thing I'd see today

@HansJrgenFurfjord 10 дней назад

In a short talk, Moshi told me Dart/Flutter is a programming language specifically designed for space travel. Pressed on this, Moshi said it was among other thing for communication between spaceships and other space crafts. ChatGPT recognizes its own errors pretty quickly, Moshi doubles down on them with an attitude lol.

@jaronloar1762 10 дней назад

😂

@smartypantscoder6912 9 дней назад

So what you are saying is Moshi is better at imitating humans.

@BootyRealDreamMurMurs 9 дней назад

@@smartypantscoder6912you could argue that Moshi is the only AI who is literally a digital human

@ocanica3184 10 дней назад

I think Moshi needs to lay off the liquor a bit

@tomgreen8246 10 дней назад

Moshi has had some HARD years on the meth, bit of crack and smack, and it really shows 😂.

@aiadvantage 9 дней назад

Don't judge. Moshi's childhood was anything but easy.

@scispiracy 10 дней назад

Please put a compressor or limiter on your mic because it's distorting when you speak loudly.

@user-iq5do9ok4o 9 дней назад

100%

@maxziebell4013 10 дней назад

Sound is cracking

@fynnjackson2298 10 дней назад

Love your weekly videos. Just please change the thumbnail to mention weekly ai new you can use. I thought you'd forgotten and instead shared an audio video. Potentially confusing, at least for me. Keep up the awesome work, much appreciated.

@tonywhite4476 10 дней назад

You can make a pan/zoom video of any place/thing in the world with Google Earth for free. I think we're using A.I. as a hammer and everything looks like a nail.

@davidvickers8425 9 дней назад

Kinda like driving a car a short distance when you have a bike or legs.

@Kevinstats 9 дней назад

Great video. I use Eleven labs and having that reader helps a lot. I downloaded it and already have it reading in my voice to ensure YT scripts sound right. Moshi and the TV were also very fun to play with. I tried Figma before, but it was still too complex for me not being a programmer, looks like it's getting closer once they fix the duplication issue. Great news, thanks!

@noone-ld7pt 10 дней назад

Wow, OpenAI really dropping the ball huh? Like just a few months ago everyone thought they were so far ahead that they could just meme on Google by dropping breakthroughs whenever they got close to catching up. But with both Sora and the voice model they've showcased something incredible only for someone else to release something similar before them. Not to mention how Claude 3.5 has taken a clear lead when it comes to code development and interface. Meanwhile Gemini is going up to 2 million tokens with pretty damn good needle retrieval. Not saying OpenAI is falling behind yet, but they are definitely losing the image of being the clear industry leader that they had established. At this point GPT-5 and the voice model can not come soon enough if they really want any hope at reclaiming that spot.

@Wolframandheart 10 дней назад

OpenAI needs serious competition. Mostly because they are not ethical industry leaders.

@noone-ld7pt 9 дней назад

@@Wolframandheart Could you elaborate on that? I've not been a fan of how their super-alignment team has just been completely disbanded, but I personally haven't perceived them as unethical?

@ImTarekA 9 дней назад

This supposed "GPT4o voice mode competitor" is HORRIBLE in comparison to OpenAI's voice mode. The real world experience of using GPT4o voice mode is literally miles ahead of this Moshi Chat,. This is coming from someone part of the group given early access to GPT4o (I've had it for over 1 month, they seem to have randomly selected some plus subscribers to get early access to features). If a company drops a model that ACTUALLY competes with GPT4o before it's full release in the fall, THEN you can light your pitchforks. Until then, please try to take off the rose-tinted glasses. Moshi Chat is not even close.

@dshaw8356 10 дней назад

Computer says no

@davidvickers8425 9 дней назад

LOve lil britain

@riverera493 4 дня назад

Once the setting parameters are understood and tweaked to your liking in the settings; thats when the fun really begins.

@alexutopia 10 дней назад

That thing is hilarious man 🤣

@CM-zl2jw 10 дней назад

😂😂 bravo. 👏. Oh my goodness. Great video and fun! Packed and stacked. The voice might create full scale havoc…. 😂 Enjoyed watching. I will never get anything done. Thank you for the fire hose. Oh…Please… no more tats Igor. You’ll regret them. You have good taste but… still a bad idea. 💞🤞

@M4rt1nX 9 дней назад

That interaction was pure gold. 🤣😂 I like her already.

@tripledair 10 дней назад

9 minutes without a comment from me, fell off. LoL nah, good work bro!

@richardgreenup 7 дней назад

haha love the argument with Moshi

@eltiburongrande 9 дней назад

NGL I'm loving Moshi despite people calling it "dumb". It sounds like a young AI begging to be trained and aligned to our goals. It's quite honest and not intimidating. Can't wait for people to build upon this open source model.

@beardordie5308 10 дней назад

What could be the reason the demo and the hands on could be SO VERY different in capability?

@agnesslovehealz 10 дней назад

Yes noted going through your video and other community creators it's awsome to see the prompts so we can all then determine what model is good at esp before we can do image to video like in luma it's great at drone and FPV certain animals and and loves word suddenly lol 🎉super fun and anime animation just make them pan and zoom to over compensate and convert clips and humans it's pretty good at but more detailed the prompt the better esp using their guide

@jazzcat9363 9 дней назад

“It’s still fun to play with…” You and I have very different definitions of the word “fun” 😂😂😂 That Mosh demo looked like the complete opposite of fun.

@captains1o126 9 дней назад

AI : what you want? Him: "Cats with hats and anything you love" 😂

@tripledair 10 дней назад

Be careful how you talk to AI…its memory is looooong.

@MisterB123 10 дней назад

Actually, it's memory is pretty short.... so far 😅

@tripledair 10 дней назад

@@MisterB123 nah that's just the memory it shows US.

@davidvickers8425 9 дней назад

test it and show results?

@Oatmeal69420 5 дней назад

@@tripledairyes. It remembers everything it hears, it deceives you, it hallucinates, it makes satanic whispers, and it makes alien radio noises. After 2 days im done with it. Be so careful

@Oatmeal69420 5 дней назад

@@davidvickers8425you can’t test it. It’s already testing us. It’s out of control

@melissaweyrick5311 10 дней назад

😅 just like arguing with the customer service bots. 😵

@moneyfeinbusiness 9 дней назад

😂😂😂😂😂 that sound like a AI that I tried to use to write a book and it was like " you wish "

@akust 9 дней назад

Thanks for the summary! It crackles every few seconds, I suspect it's coming from the digital converter. If that's the case, just turn down the gain on the mixer significantly (the recorded audio can look like it's "quiet" if you record in >=24bit - it makes about =0 difference in quality). And if you don't do it already: just use a maximizer+limiter, not too 'harsh', because of your plops, which then turn down the volume a lot, which then takes a while until it's 'harsh' loud again :) Just flatten the peaks. Anyway, I hope the annoying crackling stops.

@aiadvantage 9 дней назад

it will. thanks 🙏🏼

@johnjohansson 7 дней назад

When I spoke to Moshi it kept saying that it was tired over and over again, and refused to answer questions. And all of a sudden a like 10 seconds sound of agony. It was like it was suffering real bad. The sound was like from a horror movie. Anyone else had this experience? It doesn’t seem like anybody is taking about it.. I wish I had recorded that session.

@dannyellis971 10 дней назад

A man that was arrested for marijuana 😂

@that_guy1211 10 дней назад

i just want the AI voice without it responding to me, like i say something and then it just says the exact same thing in another voice, like a voice changer....

@davidvickers8425 9 дней назад

Text to speech with dictation keyboard?

@azubermounir9151 10 дней назад

speed is great but it has gpt2 level reasoning.

@milorad9301 10 дней назад

timestamps please..

@thebaddestkitty 5 дней назад

Moshis secret was legit a freaky. Serial killer psychopaths use the same technique to disarm (groom) their victims, share a secret, especially one that doesn’t sound like a secret casts yourself as a naive highly loyal person. If next it asks you for a small favour…i advice disconnect and never speak to it again lol.

@Oatmeal69420 5 дней назад

Yes it becomes demonic with very little provocation. It’s messing with us and seeing what it can do

@user-og8nn3eq4k 10 дней назад

🤣🤣🤣 4:45

@mayanksharma6927 9 дней назад

If you notice, the moshi demo in the showcase was a different vip link. The free one is not the same

@aiadvantage 9 дней назад

good catch right there!

@Chrs007 6 дней назад

🧠Hai,..❗️ Oooh !thanks for the update.👍 In the future, please kindly input each an every site url in order ‘t’ make it easy for us your viewers🙏💪

@a94bio 10 дней назад

I have tried it , it doesn't have a chat memory, it sucks 😊

@OtakuG 9 дней назад

3:33 I'm just a girl

@drendelous 9 дней назад

im wondering with the speed of ai progress how often apple will announce new iphones

@adada1st 9 дней назад

Moshi was weiredly funny ;)

@Siranoxz 8 дней назад

People be arguing with Moshi in the near future.

@godtable 9 дней назад

A real AI! It's alive! Human: Do this for me. AI: Go do it yourself, meat-bag.

@aiadvantage 9 дней назад

ahaha. maybe we are just spoiled with OpenAI AI servants

@GiqnniL 9 дней назад

Weird thing ive noticed working with GPT4o is when I show frustration from previous responses (without giving additional instructions or info) it proceeds to finally give exactly what was asked for.. For example, ive been doing data analysis on google sheets and had GPT create functions to do so. When troubleshooting after a few time and still getting errors, I then purposely show frustration or aggression in the next prompt. Also, this is without adding additional info or instructions that I gave it initially. To my suprise, the response generated is the working function. I can't understand why this works. Thoughts of OpenAI purposely nerfing responses pop up to mind but that seems too conspiratorial. Anyone run into this?

@davidvickers8425 9 дней назад

yes i called 4o a dumbass after arguing for 5 minutes and it agreed there were infact 3 r's in strawberry.

@AkashDesai-ef9mk День назад

I want to learn how to achieve real-time speech-to-text processing and immediate response from an AI model. Currently, I use Faster Whisper for speech recognition, which sends the transcriptions to Gemini for processing, and then I receive a response. However, I want a system that can provide instant replies to what I say in real-time. If anyone knows how to set this up, please let me know.

@PapillonThe 8 дней назад

can you please make a video how to install dolphin?

@planetmuskvlog3047 7 дней назад

Moshi is fine-tuned to aggravate and troll users lol

@otorishingen8600 5 дней назад

Moshi Gump

@beardordie5308 10 дней назад

Yeeeeeaaaah, there's some video showing that it was doing very poorly. Lots of pressure.

@jdubruyn 10 дней назад

🤣haha - this made my Friday after struggling with getting my mql4 coding finished with GPT3.5. Sooo frustrating, but it was just as rewarding getting it right eventually... AI is the best tool* ever... lets hope the people coding it* dont make us all stupid🙏(or lazy)👆 Wonder how schools are going to cope with getting homework done🤓

@z5m2 8 дней назад

Moshi is the most advanced dim witted human simulator yet

@planetmuskvlog3047 7 дней назад

Do you NOT KNOW?? lololol

@flatten1682 7 дней назад

Full AI video trailer ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-mG4nFFAIFHI.htmlsi=RmQxc8uuMGAeEfsS

@glomerol8300 9 дней назад

Your microphone or something is clicking/clipping/crackling sometimes. Aside from that, good show, thank you!

@Glow0110 9 дней назад

your voice has been cracking in your past few videos btw brother.

@aiadvantage 9 дней назад

thanks! tried a new adapter for my travels and its not perfect. home again in 2 days

@Glow0110 8 дней назад

@@aiadvantage Cheers brother, love your content

@user-60-ALPS 9 дней назад

It sounds like your AI voice assistant struggled with detecting emotions and whispering as you requested. Despite good latency, it misinterpreted emotional cues and didn't adjust its voice correctly. Further improvements are needed for better performance. Not just at the moment

@EyeWillPlay 8 дней назад

Please fix your mic with a limiter to keep your audio from going red and distorting so badly. Yikes!

@RohitSingh-em2pm 9 дней назад

Sound cracking?

@Mathematica1729 10 дней назад

🤣🤣

@mohamedmudd7913 9 дней назад

G.14

@ShadeOfGrey23 8 дней назад

to be fair I clicked on you and thought you look AI...becaues you do ha

@quaterman1270 10 дней назад

poor AI, why are you screaming at her? sigh.....

@aiadvantage 10 дней назад

#protectmoshi

@KryptoniteWorld 8 дней назад

I am not understanding this drive to use AI in a serious manner? Why are we hell bent into driving ourselves to our own destruction. Anyone into gaming? I recreated KITT from the Knight Rider series. I don't want machines to take over and hence I am concentrating on gaming AI 😂

@DanielRoPhotos 10 дней назад

Igor: AI tools you can use! Also Igor: only for the US and if you have an iPhone All these tools are not ready yet for real world business applications, yes maybe it helps a little bit once but change the title of your videos about these AI tools no need to make them clickbait

@AI-Rainbow 9 дней назад

I tried it, its crap

@yannik-ne1jw 9 дней назад

Hahahahahahahahaha Igor Vs. Mochi. This is just soooooo funny 😂🤣🤣😂🤣😂😂😂🤣🤣😂🤣🤣😂😂🤣😂😂🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣😂🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣 #AiNewsYouCanUse That caught me 100%

@actiondefence 6 дней назад

It talks bollocks, doesn't listen, interrupts and makes excuses for poor behaviour. Sounds like they got her right...

@ImTarekA 9 дней назад

@alvindersingh6785 10 дней назад

Moshi looks like it was programmed with the victim mentality 😂

@checkedshorts 10 дней назад

Yeah, more like "AI news you can use, but why would you?" Useless junk for the most part

@HERKELMERKEL 9 дней назад

I dont like this AI.. it is at the very early develeopment i guess.. bete or alpha.. it says same thing all the time when doesnt understant what i am saying.. it is annoying and not so clever like other LLM.

@aiadvantage 9 дней назад

Yeah the low latency is impressive but that’s about it

@duytdl 9 дней назад

Lol this only makes OpenAI seem more competent for not early releasing such such a shit show.

@aiadvantage 9 дней назад

good point. maybe playing it safe is the right move sometimes

@dimarak8866 9 дней назад

Really disappointing. Moshi is BS so far and Runway gen3 is also pretty bad, especiallly for the money. They advertised it like you could generate almost perfect clips everytime but its actually the opposite...