Mysterious AI Music Generator 2x Better Than Suno?!

so over the last few days there have been some whispers about a new generative AI music platform that is better than sunno naturally I decided to pop on my detective's hat and see what it was all about and I think I've collected pretty much everything there is to know about this new Challenger to the throne not only will we hear what it sounds like but who it is and as a bonus I've got a brief chat with like a legit Rockstar and professional musician about AI okay let's dive in first off I do have to say that some of this story takes place in the Shadows I tried to do as much like legitimate reporting as I could like in my sore Deep dive recently but due to the nature of ndas obviously some people are just going to be referred to as sources I on the other hand can talk about this because no one asked me to sign an NDA so this whole thing kicked off via a tweet from mayay Wrigley who said just got a demo of an AI music product that's two times better than sunno most Flor I've been since Sora demos the model is absolutely wild AI music will soon take the World by storm sire then responded with probably what all of us were thinking so what's the point of this tweet then at that point like a pirate ship made out of Rice Krispy Treats the leaks started coming a few days after McKay's original tweet Max posted I got access to an upcoming AI music app and am mind blown by the outputs it's like Sora but for music Jimmy apples also responded to McKay's original tweet stating if it's the one I'm thinking of then yes it made sunna V3 sound me in comparison no small feet now you'll note up to this point there is no mention of who this mysterious puno competitor is nor have we actually heard anything but here's where things get interesting because mumai uncovered a now deleted post by XG project featuring an AI generated song with a animated video background that we have not seen yet the song generated is very much like golden age of Hip Hop Moon byai refers to it as Tupac to my ears it sounds a little more like Tribe Called Quest but let's take a you let me know who you hear in the world of dat where the details there's acees and strength to show through rows and columns it weaves it Quest and algorithm Stand Tall Above the Rest XG boost they call it with power so blast in the race of machines it's unsurpassed right the gradient soaring high ex bner across the sky think you need in a dance so B find it P precision Embrace we TR theis got another leak sample coming up in just a minute but for now let's dig into this one first off the song is very good I'm not hearing any weird artifacting there's a little bit of flanging on the high hats at the very beginning but that was actually something that was fairly common in tracks of that era the song does stay in key and isn't like randomly transposing itself although to be fair that was more of AO like V1 and V2 issue and in terms of instrumentation rhyme and flow the fact that it has two MC's everything does sound really really good there's also that Sly guitar hook which is definitely something that you heard in '90s alternative hip-hop additionally we can see here that it is generating in Stereo which you know qualifies at least to some degree the claim that it is two times better than sunno in terms of overall Fidelity it's kind of hard to tell we're listening to something that was screen recorded uploaded to Twitter compressed and now uh recompressed as I present it to you on YouTube I can tell you that loading it up into an EQ there is like this really hard sloping cut uh somewhere around like 12K that probably is the Twitter compression similarly with the track that we will be hearing shortly again we also are getting that hard cut off around 12K so I think that this is definitely more of a compression issue than the model backing that up with one anonymous source the sound quality is a lot better on platform and via another anonymous source the sound sample that we just heard isn't even quote one of the good ones now for the fun part who who is this well swisser fever was able to sus out that the platform's name is udio now I don't know if they're dropping the a in audio or the St in studio but if it is udio as in studio then that's awesome because it totally reminds me of the classic delirium sequence from coffee and cigarettes this was a 2003 film by Jim jarmo kind of a slice of Life thing but the delirium sequence was Rizza jizza and Bill Murray hanging out and you know drinking a bunch of coffee and herbal tea it's wonderful and if you haven't seen it it's linked below and in terms of confirmation that it is udio we do notice that there is this kind of like triangle Corner which does appear also in the video generations to further confirm legit rumors went and looked at the web code of udio and discovered uh the options that we will eventually have here obviously you will be able to create a song via a prompt under artist Styles it is noted that we do not generate songs using artist's voice but you can use artist styles to describe music and we will replace the artist reference with relevant genre tags this is similar to sunno in that you can't prompt specifically for an artist name but you can prompt for keyw wordss that are relatively associated with that artist and get something that kind of sounds in the ballpark we'll obviously be able to write lyrics and have descriptors like chorus hook verse uh it cuts off there we may have other options remix which will allow you to create new tracks inspired by songs you like or create variations to tracks that you or others have created now an important note here is that I do not think that you will be able to audio reference with this as you can with stable audio too like you're not going to be able to upload an MP3 of a pre-existing track and have it spit out something that sounds like that track additionally legit posted up a second sample this one from an anonymous source kind of more in a soul Vibe and a much longer generation uh the vocals on this kind of blew me away let's take a listen guess it's time to Swit that AI ride got to face the music dwell on the P building up the rym going to make it last Alex oh we had our time but I'm moving on I'm going to find my Prime I can't believe they ner the No No a [Music] train I can't believe they ner the no s to mym I can't now I'm going to start my work oh Alex baby come on now listen friends it's time to subscribe to Lady space part I can't my I can't Dr Dr can't trust you oh Alex baby come on now listen friends it's time to subscribe to latent space pop so that is pretty impressive the vocals do sound very well balanced and mixed in uh occasionally with sunno I always get the feeling that there's sort of a track underneath it and then vocals layered on top this feels very much baked in additionally in that drum fill drum solo section when you zoom in on the waveform you can see that the left and right channels are definitely doing something different so yeah I mean it's it's it's independent stereo tracks so obviously udio is in like super closed beta I had to go digging to even find out who were but it does appear that if you go to the website now you can actually sign up for the weit list as a bit of a tag to all this generative AI has obviously been a bit of a controversial subject in the artist realm but in my opinion the group that seems least fearful of it and most adaptive are musicians to that I thought I'd share a quick clip from a chat that I had with Jordan rudis the virtuo musician from the band Dream Theater if you aren't familiar with Jordan's work or aren't a fan of prag rock which I totally understand it's the nerdiest of all rocks uh you might want to check out a really cool video that he recently did with a piano in which he ends up writing the piano track to uh Jay-Z and Alicia Keys Empire State of Mind without ever having heard the track anyhow here's a take on AI by a professional musician who has played everywhere from like Madison Square Garden to the budon so clearly you are you do not fear ji as not at all I actually think that it's my help for you know thinking in it is that I'm using it personally everything I do it visually and using musically is to enhance do I I mean an example going way back this on 8 bit computers and I got my first Daw okay and I was playing and I discovered quantising even the fact you could play something on a med keyboard and then speed it up and when I did that that's amazing but my first s was how let me learn how to playn how to try to make that sound my fingers CU I wanted to feel what it was like to the experience of making that kind of sound with that kind of head space is was opening up a new CH for me I never thought about it that way so for me the technology is all about opening up those other Pathways the things he Liv witha for sure and then to bring them into my own being to become you know that much uh more expressive more technical myself May me a little wiser and just have an an opening yeah yeah I'll keep you updated on what's going on with udio in the meantime I thank you for watching my name is Tim

Channel: Theoretically Media

Views: 52,536

Keywords: ai music, ai music generator, suno ai, suno ai music, ai generated music, ai news, ai music production, ai tools, ai music tutorial, suno v3, ai song generator, ai song

Id: 0kU4bGpEtWU

Length: 10min 7sec (607 seconds)

Published: Mon Apr 08 2024