Noise removal has been around for ages in different audio software, but the leaf blower example without the voice sounding like a total robot is insane.
Yes but (there's always a "yes but" 😄 ) the leaf blower has very consistent noise. It'd be more challenging to have background noise that is random - how well does it do then.
Gen 3 seems more stylized while Sora seems more photorealistic. I don't think Gen 3 is any worse, but Sora behind the scenes is probably much improved from its reveal. We could get another Sora reveal before it finally releases.
after using it 3 days straight sense release 10 hour shifts i can say its pretty good with a lot of things but very bad with some things like feet,hands,unpractical stuff ect
Runway Gen3 10 second videos use up 100 credits, which on a starter subscription means you get 6 per month. I really think this needs to be shared because it stung me, and I can't be alone in that.
@@ZerofulMaster I don't know for people on USA, but anywhere else this is too expensive. Here in Brazil, for example, that $95 would bel R$518,72 (reals). The minimum wage in Brazil is R$1.412. Unless you work with this or can convert those videos in money, it's not worth it.
@@ZerofulMaster People complaining about that $95 bucks a month but you have people complaining that they spent $4k on Luma, and devs on Luma's Discord going on about how they have to charge high rates because of this and that when Luma just got a ton of seed money. Just greed
HUGE Thanks to Invideo AI For Sponsoring Today's Video! Go to: invideo.io/i/mattvidproai MATTVIDPROAI50 (For twice the number of video generation minutes in the first month)
Adobe did a free AI voice filter like that 1 year ago, and I think it's still better. It doesn't just remove the background noise, but it reconstruct the voice wave. I fixed some very badly recorded telephove calls from the 80s with that, it's incredible, you should try it.
We karaoke creators have known how to do the vocal isolation for years. Split the voice like you’re devocalising, but keep the voice instead of the “music”.
If I had to take a shot for every time OpenAI has been late in delivering an AI product they teased ages ago, I'm not sure I would be able to stand up (for Google, I might be dead though).
@@BackTiVi OpenAI still takes the cake against Google. OpenAI has said something about CriticGPT, Gpt4O voice, gpt4o vision, sora, and havent done anything
Nice one,Matt, cheers for the update! Hey, how about another Moshi conversation? That was the funniest thing I've seen in ages man. I know what you're gonna say.... "I can't. I can't. I can't." But you CAN! :)
Hey there! I really enjoyed the video, but felt that the sponsor segment was a bit lengthy. Maybe it could be shorter next time? Thanks for the great content though! 😊
A million dollars per LLC. Glad that my Stability issue has stabilized. I just hope as they continue to grow they "keep" listening. We have enough Adobes and Oracles already. GenAu sounds good enough for layered bg audio ;)
Here's some I've been keeping up with. Maybe these could help you or someone viewing these comments Mattvidpro Matt Wolfe TheAIGrid AI Explained Matthew Berman David Shapiro Dylan Curious AI Advantage TwoMinutePapers (what a time to be alive!) ComputerPhile TheAiBreakdown Fireship (not fully AI focused but I do get useful news from them) Bycloud (same as the previous one too, useful but not entirely AI)
have you tested sunos new audio feature do you think its better the udios i tested it and just by entering the word "continue" it turned this jazz clip into a full edm song lol
Udio will not listen to the damn prompt , it will never change genres, only keep continuing the track. Tbh Udio is way less consistent than Suno and spits out WAY more garbage regardless of being a continuation or not.
So I built one of those AI meeting assistants that sits in meetings and makes notes, tasks action items and such. Started 2 days ago and I've never built a react app before. The better you are at using these tools the more you can do. I simply cant imagine what I could accomplish with Opus 3.5 Right now I think I could get to a decent personal assistant/co worker using a home server.
Krea "scene transfer" is just ip adapter + relight , these websites arent creating anything they are implementing comfyui workflows that have been out for months and then act like they created these features . Meta 3d Gen didnt release anything except for a paper and renders , no code , no weights , it is not "released"
WTH they will realease a better version of stable diffusion 3 AND finally gave a good license so we can finetune and build loras and so on??? heell yeah! so in a few months we should see some community trained models then :) NSFW please? we should find the best people who contributed to sdxl finetunes (pony and juggernaut) and crowdfund gpu for them!
@MattVidPro the leaf blower has very consistent noise - a more interesting, more telling demo would be on random noise, like a crowd in the background or crashing ocean waves, etc
I hope seriously an idea better than now world, that everything is free to access and use and pay to get paid using it, or pay to use it extensively. Or to unlock certian things. I don't even like this though. Everything should be free. And anyone with ideas should be given attention and support and resources. And nobody should work for money. Just my opinion...
Yes nasty reaction to any stupid "feature" that thinks trading all your privacy is worth something we have never needed - All capture-all-your-data features should be local open source software completely under your control only. If I want to place all my photos in a directory on my computer and have a local AI inquire on them then that's my prerogative. It should not be pre-implemented. Do not ever promote such things
Are you, like, giving him an order or something? “Do not ever do this.” Yeahhh ok. I know I personally always follow the instructions of randos on the youtube comment section. 🙄
@@CaritasGothKaraoke no I’m telling people in general and him as well to never promote such a horrible feature. It’s dangerous. So that requires emphatic feelings about it.
@@apache937 Your device is not local only though. You are still trusting google. This vs a PC at home not connected to the internet handling the AI ops.
You should first very quickly clearly explain any terms that might be unfamiliar to your viewers. I had to figure out what you meant by "Video Outpainting" - your explanation was incomplete
Im so tired of AI content creators bringing up SORA, it WILL NEVER! be released. It massively over-promised/lied, each SORA video took a massive compute budget to create, its not viable for a mass market. Its why Kling, Gen3, and Luma are all massively downscaled versions of Sora.
Genuinely, thank you for finally consistently adding chapter markers.
Noise removal has been around for ages in different audio software, but the leaf blower example without the voice sounding like a total robot is insane.
Yes but (there's always a "yes but" 😄 ) the leaf blower has very consistent noise. It'd be more challenging to have background noise that is random - how well does it do then.
I wonder if they're using a form of voice-cloning to supplement the recorded data.
Gen 3 seems more stylized while Sora seems more photorealistic. I don't think Gen 3 is any worse, but Sora behind the scenes is probably much improved from its reveal. We could get another Sora reveal before it finally releases.
after using it 3 days straight sense release 10 hour shifts i can say its pretty good with a lot of things but very bad with some things like feet,hands,unpractical stuff ect
Sora ain't out yet. We can make up anything we like about it, but it isn't out so it's all just speculation.
definitely we need a KREA Style Transfer video.
Runway Gen3 10 second videos use up 100 credits, which on a starter subscription means you get 6 per month. I really think this needs to be shared because it stung me, and I can't be alone in that.
I've seen many promising AIs fall because of that. It doesn't matter how great it is, if people can't pay it, it's useless.
@@End-phoenix Completely agree
You can buy unlimited generations for 95$ monthly
@@ZerofulMaster I don't know for people on USA, but anywhere else this is too expensive. Here in Brazil, for example, that $95 would bel R$518,72 (reals).
The minimum wage in Brazil is R$1.412. Unless you work with this or can convert those videos in money, it's not worth it.
@@ZerofulMaster People complaining about that $95 bucks a month but you have people complaining that they spent $4k on Luma, and devs on Luma's Discord going on about how they have to charge high rates because of this and that when Luma just got a ton of seed money. Just greed
HUGE Thanks to Invideo AI For Sponsoring Today's Video!
Go to: invideo.io/i/mattvidproai MATTVIDPROAI50 (For twice the number of video generation minutes in the first month)
👋
Magnific has a feature like Krea has too.
Also, we have to be skeptical of sora until they actually release something to test. The Ballon head video was made along with editors
Adobe did a free AI voice filter like that 1 year ago, and I think it's still better.
It doesn't just remove the background noise, but it reconstruct the voice wave.
I fixed some very badly recorded telephove calls from the 80s with that, it's incredible, you should try it.
We karaoke creators have known how to do the vocal isolation for years. Split the voice like you’re devocalising, but keep the voice instead of the “music”.
I'm a bit surprised you didn't mention live portrait. Great new open source portrait animation tool.
Can anyone even use Sora yet? Feels like they dropped the teasers ages ago
If I had to take a shot for every time OpenAI has been late in delivering an AI product they teased ages ago, I'm not sure I would be able to stand up (for Google, I might be dead though).
@@BackTiVi OpenAI still takes the cake against Google. OpenAI has said something about CriticGPT, Gpt4O voice, gpt4o vision, sora, and havent done anything
Nice one,Matt, cheers for the update!
Hey, how about another Moshi conversation? That was the funniest thing I've seen in ages man. I know what you're gonna say.... "I can't. I can't. I can't." But you CAN! :)
I’ve been wondering how much of its confusion came from Matt calling it “Mosh E” instead of “Moe Shee” like ot should be.
Hey there! I really enjoyed the video, but felt that the sponsor segment was a bit lengthy. Maybe it could be shorter next time? Thanks for the great content though! 😊
6:13 So about 66% of rode island state (3,144 km2). Now you just need the hydrogravity batteries to store the power for night use.
A million dollars per LLC. Glad that my Stability issue has stabilized. I just hope as they continue to grow they "keep" listening. We have enough Adobes and Oracles already. GenAu sounds good enough for layered bg audio ;)
You are one of the 5 Best channels talking about AI.
Then please mention those 4 other channels. I was searching for this level quality in other youtubers, it would be helpful to me.
@@BrittoMaryChannel called ”AI Explained ”. Its a bit more advanced tho
@@BrittoMary Matt Wolfe, Matthew Bermann, Dylan Curious, David Shapiro, The AIGrid
Here's some I've been keeping up with.
Maybe these could help you or someone viewing these comments
Mattvidpro
Matt Wolfe
TheAIGrid
AI Explained
Matthew Berman
David Shapiro
Dylan Curious
AI Advantage
TwoMinutePapers (what a time to be alive!)
ComputerPhile
TheAiBreakdown
Fireship (not fully AI focused but I do get useful news from them)
Bycloud (same as the previous one too, useful but not entirely AI)
@@BrittoMaryaigrid
perplexity was started to kill the google search engine
Hello cool video. I hope you had a great 4th of July.
have you tested sunos new audio feature do you think its better the udios i tested it and just by entering the word "continue" it turned this jazz clip into a full edm song lol
Bruh what prompt did you put
@@Insertrandomnamehere412 i literally just put the prompt "continue" and for some reason it made edm lol
@levonkenney you could of put jazz on the prompt atleast
Udio will not listen to the damn prompt , it will never change genres, only keep continuing the track. Tbh Udio is way less consistent than Suno and spits out WAY more garbage regardless of being a continuation or not.
@funkahontas Nah I had far more better tracks in udio than suno. Most of them are rock music and guitar solos
Great video and thanks for the chapters!
So I built one of those AI meeting assistants that sits in meetings and makes notes, tasks action items and such. Started 2 days ago and I've never built a react app before.
The better you are at using these tools the more you can do. I simply cant imagine what I could accomplish with Opus 3.5
Right now I think I could get to a decent personal assistant/co worker using a home server.
Any idea you see, can be parsed into sections and built using Sonnet 3.5
Great videoo - please make a video on scene transfr
They must have some Sora 2 candidates ready by now
I like what stability AI did there.
The model we can use is the best model.
Krea "scene transfer" is just ip adapter + relight , these websites arent creating anything they are implementing comfyui workflows that have been out for months and then act like they created these features . Meta 3d Gen didnt release anything except for a paper and renders , no code , no weights , it is not "released"
I can't scan the QR code, I'm watching this on my phone. 😂
I guess that's why there's a link.
Music video using Krea AI Style Transfer:
ua-cam.com/video/ypHnqh3ICCo/v-deo.htmlsi=OmMrLu65RFJVR1Ak
The excuse of I am not a lawyer doesnt travel as far these days... Latest AI is pretty damn good so just run it through
Tried Runway for putting motion to my Photoshop edited Ideogram / Midjourney images for use in my restaurants.. aaand useless. Waiting it out further
I’m here for the crazy hair 💇♂️💇
WTH they will realease a better version of stable diffusion 3 AND finally gave a good license so we can finetune and build loras and so on??? heell yeah! so in a few months we should see some community trained models then :) NSFW please? we should find the best people who contributed to sdxl finetunes (pony and juggernaut) and crowdfund gpu for them!
i want to ask you something, do you have another youtube channel? i think it's name was E-Q something 😊
Now back to the video 🚫
back to your regularly scheduled content ✅
That moment when you're not blown away this year on AI advancements, welcome to the start of it's plateau
You are the best AI UA-camr who talks just about AI news who agrees with me 👇🏻:
mee!!! 🎉
Agreed
voice isolator ... ever heard of RTX voice? that's how old now
That's what I thought, I use RTX audio on nVidia and AMD Noise Suppression on Radeon cards for ages now. :)
This is def better tho
@MattVidPro the leaf blower has very consistent noise - a more interesting, more telling demo would be on random noise, like a crowd in the background or crashing ocean waves, etc
I hope seriously an idea better than now world, that everything is free to access and use and pay to get paid using it, or pay to use it extensively. Or to unlock certian things. I don't even like this though. Everything should be free. And anyone with ideas should be given attention and support and resources. And nobody should work for money. Just my opinion...
Pls pronounce prosche like porchia xD
Yes nasty reaction to any stupid "feature" that thinks trading all your privacy is worth something we have never needed - All capture-all-your-data features should be local open source software completely under your control only. If I want to place all my photos in a directory on my computer and have a local AI inquire on them then that's my prerogative. It should not be pre-implemented. Do not ever promote such things
Are you, like, giving him an order or something? “Do not ever do this.” Yeahhh ok. I know I personally always follow the instructions of randos on the youtube comment section. 🙄
@@CaritasGothKaraoke no I’m telling people in general and him as well to never promote such a horrible feature. It’s dangerous. So that requires emphatic feelings about it.
is the pixel ai local? if yes then its ok. it doesnt take screenshots, just uses ones you already took
@@apache937 Your device is not local only though. You are still trusting google. This vs a PC at home not connected to the internet handling the AI ops.
@@gregblank247 this is the case regardless of not your phone has any ai features
Firififfjrjjcifiee
Not local? Not real.
You should first very quickly clearly explain any terms that might be unfamiliar to your viewers. I had to figure out what you meant by "Video Outpainting" - your explanation was incomplete
If you’re so much better at content creation, where’s your uber-successful channel, mate?
@@CaritasGothKaraokeit was a suggestion indicated point blank. It’s Feedback with the intention to help him
Dude if you don't already know day-1 terminology then you might be on the wrong channel
@@acain6803i didn’t know that this channel didn’t welcome people who don’t know each and every single bit of new terminology.
thats obvious if you have beenn in the ai space for a bit. this isnt some toddlers toy channel
Im so tired of AI content creators bringing up SORA, it WILL NEVER! be released. It massively over-promised/lied, each SORA video took a massive compute budget to create, its not viable for a mass market. Its why Kling, Gen3, and Luma are all massively downscaled versions of Sora.