This AI is a dream for music production
Вставка
- Опубліковано 6 лип 2024
- This AI feature is a game changer for musicians! #udiomusic #aisongs #udio #suno #aimusic #ainews #ai #agi #singularity
Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: chatllm.abacus.ai/?token=aisearch
Newsletter: aisearch.substack.com/
Find AI tools & jobs: ai-search.io/
Donate: ko-fi.com/aisearch
Here's my equipment, in case you're wondering:
GPU: RTX 4080 amzn.to/3OCOJ8e
Mouse/Keyboard: ALOGIC Echelon bit.ly/alogic-echelon
Mic: Shure SM7B amzn.to/3DErjt1
Audio interface: Scarlett Solo amzn.to/3qELMeu
CPU: i9 11900K amzn.to/3KmYs0b
0:00 Intro and examples
3:53 The problem with AI music generators
6:22 Riff to full song
12:29 ChatLLM by Abacus
14:02 Extending the song
20:41 Singing to full song
24:49 My song to female voice
27:00 My song to duet
28:28 Instrumental to full song
31:14 Instrumental to instrumental
32:24 Nokia ringtone to instrumental
34:18 Nokia to full song
35:39 Riff to J-pop - Наука та технологія
Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: chatllm.abacus.ai/?token=aisearch
👋
🖐👋🖐
LOL
I was so surprised by the windows XP theme continuation and orchestration 😮
it was so beautiful
Yes that stunned me
@@theAIsearchreminds me of you
Yeah it turned it into a Jurassic park Background score.
It was Excellent!
The vocals added to your singing are so angelic
It almost masks the cringe singing but makes it a part of the song
lol
@@staticlee4287they need a feature to get rid of your own voice. I'm a songwriter who has a terrible voice
@@theAIsearchcool
@@theAIsearchI can’t tell if it’s an insult or a compliment. 😂
As I watch Ai unfold I keep finding myself sitting in disbelief, mouth open, what a tool to work with, and this is just the beginning. And yes I have subscribed!
Thanks for the sub!
@@theAIsearchyoure welcome
@@theAIsearch🤭 He may have meant to UDIO?
No freaking way
The nokia guitar theme has never been so different, full and mind blowingly real 😮
The wxp start up sound is even better 😂
@@TheSoundChronicles the xp one was so beautiful
yes the most amazing
@@theAIsearchlike you
thanks for not rushing this video out!
you're welcome!
The last one was so beautiful
I’m mind blown by this, wow! Thank you for the demonstration! 🙌🏼🔥
My pleasure!
Jaw dropping.
man the Ai version of the Windows XP startup at 3:24 sounds really nice
dude it sounded like windows xp trailer music or somethin
@@ItzSpecky Fr
It feels so soothing for no reason
@Khann_2102 Udio V1 audio is much more solemn and elegant than that of Suno. I hope that doesn't change in future versions, because is the soul of that AI haha
*Dude that wasn't cringe you actually sound way better than me i have to use alot of effects to just even get my voice to sound decent if ever want to sing*
thanks 🥲
Wouldn't it be better if they just allowed us to upload an audio sample of our own voices and trained it to make songs with our own voices?! That would be insane.
that would be nice. a workaround would be to generate a song, then swap out the voice with your own using RVC ua-cam.com/video/ixB9oalT3cQ/v-deo.html
That would be great additional feature, that you could apply instruments as well. It would be great if this expanded with multitracks for additional remixing, say 8 tracks.
And in their legal contract make sure that what you upload becomes their property so that you hand over the right to yourself when you do 😉 Yep, that is probably their next step.
I think Suno AI allows that. You can give it a shot.
Intriguing point .. I however, am not that attached to the sound of my own voice lol
I just can't stop commenting XD this is exciting stuff. 👍Subscribed👍
The first song actually SLAPS!! I need it in stremaing or YT or soundcloud!
Why don't you ask him directly. He could hook it up for you
Songwriter here. FWIW, it's not simply a matter of syllables if you want to get things to match up musically. Train yourself to think in terms of meter. You may have words that have the stress on the first syllable, others on the second, and only sound natural if they're sung like that. Then there are individual words and the role they play in the phrase. With rare exceptions, you don't normally want the verbal emphasis on an article (a, an, the) or conjunction (and, or). If you're unsure, speak your line out loud and try to hear the natural rhythm of its phrasing.
Here's an example of two lines to compare:
"River flowing to the sea." Starting with the first syllable, the meter is alternating, emphasis on, emphasis off throughout the line. There are four "strong" syllables. (ONEtwoONEtwoONEtwoONE)
"Reverse the engineering." Here the meter also alternates, but it is one syllable off compared to the "River" line, because the first stressed syllable in this line actually comes second. There are only three "strong" syllables. (oneTWOoneTWOoneTWOone)
So even though there are the same amount of syllables, in order to get the lines to match up, the AI will have to rush, extend, drag, or drop something in order to get the lines to actually match musically/rhythmically - or else it will have to provide an unnatural pronunciation, which is likely even worse. (Extension can of course sound great, but it may be difficult to get the AI to actually give you something that sounds good.)
This is great to get a heads up on the audio upload function. I still think the audio fidelity from Udio just isn't there yet, but as soon as they can get it to CD quality... Wowsa.
i'm looking forward to that!
Looking forward to music upscaling too!
This is actually amazing for producers who already learned music theoryband all the things that go with making music. We get to skip the blank page and have some inspiration on where tonnage the track. I love this.
Mewboesbwont make realistic music NC they lack the 10,000 hrs it takes to learn music from scratch.
This is a tool and a time saver
Thank you , very nice job..! I want ask you: if I upload my voice that is not always in tune, will the AI correct it?
the windows xp startup sound extended by ai sounds like the Windows XP Tour music
That was not cringe, your song was really good and your voice is cute, and BY GOD, I AM HEARING AS I RIGHT, WTF!? THE AI ADDON IS SO DAAAAMN GOOOD!!!! O FUCK, I HAVE NO WORDS....
thanks!
@@theAIsearchi think she likes you maybe
"CUTE" LOL! You're very kind
Awesome video! Thanks for all your hard work and insights!!
my pleasure!
Hi, do you know how to change the names of the songs produced in Udio? Thanks!
Thanx for the sharing. I am a former composer who have a lot of old tracks from 1990's that would like to enhance and clean with AI. What software would you advise using?Thanx. Peace & love
that singing was so good lol! i subscribed im looking forward to more of ur videos ❤
thanks!
At 3:20 is insanely good. Can you post the Udio link to the Microsoft XP extended startup? When Udio is trained with the NVIDIA B100 it will be game over.
yeah i really loved that one. i'm trying to find the source now!
found it! www.udio.com/songs/8ACm7odbDDyuzEHyfJDexC
@@theAIsearch Thank you. I forgot to mention your vocals and lyrics are very good. I have not had success with this new Udio feature, I tried to extend the back to the future theme song, the results were not impressive.
@@cortezforever Thanks! It's still very inconsistent and I find that only 10-20% of the generations are actually decent
Is there way to 'save' a generated voice that you really like? Perhaps if you want to make lots of songs with the same voice etc?
what a fantastic video - thanks so much for you hard work
My pleasure!
Hi, are the generations private? because what if you use something that you created like you did and you want to protect it?
The generations are private yeah, unless you want to make it public
Any idea if it will give you the different stems for the output? For me this would be useless in one big WAV format. It would be extremely useful if I can get the different stems for the different parts that it’s outputting / generating.
You can use a stemming model like demucs for that. Won't be perfect but pretty good.
Be aware that Udio has the right to use your lyrics and whatever audio you upload as training data or to generate music for others. It is in their terms of service.
BTW, the Nokia ringtone is from a Waltz, composed by Francisco Tárrega in 1902. It became public domain right before Nokia started to use it as a ringtone.
Good to know. Thanks for sharing!
That's why music producers should wait for open source, local, off-line AI solutions. They surely will come soon.
copyright is dead buddy, no matter where you put your lyrics and audio, as long as they are on the internet, Udio has the [right] to use them as training data
@@CodexPermutatio there is audiocraft and musicgenai but they are not capable of what suno or udio is
@@seed_92 I know. But eventually some open source capable alternative will arise.
This sounds really damn good. But I have to say I think if I were to use this personally, it would feel a bit unsatisfying. The same feeling you get when you make a track that you like but then realize that all the parts you like are just loops from Splice and you didn't really make any of it yourself. It's a very hollow feeling.
Yea, I agree that using Udio at first had an even more hollow feeling than using loops cause you at least are making a few creative decisions of which loops work together. But I see this update as an improvement.
If you have enough music production skills to create a full decent quality song already then starting off by uploading your own riff will give you more creative control of the song. You can use Udio for ideas, but still add or remove parts from there to your own liking. Kinda like Udio is a co-writer/ co-producer.
@@Will_Bruce This is true but I think you might end up in the same place eventually since Udio is trained on the greatest music ever made from the most popular artists in the world it's almost inevitable that it's share of the ideas will be the best ones leaving you with the same feeling.
Industry producers won’t care, they’ll land placements with these
@@oscard9429 Completely agree, and I'm open to using it myself. But it seems like producers these days will do anything it takes to get a placement and make some money.
damn that singing was acc fire 🔥
🤯
crazy idea but awesome results :) great thing u did!
thanks!
On the free plan I tried cropping my first 32 seconds of the song back to a certain point and adding the next section from there but that crop tool gave me inconsistent results. In the end I just added the next section without cropping.
Was a Cakewalk user since DOS days (yes, they go back that far). Now, Tracktion Waveform just runs circles around it for layout and workflow ease. It just got convoluted as they threw the kitchen sink at it over the years before going bust and getting resurrected. I use the paid version a lot but also have Reason and Bitwig.
Once we fix the random generation issue and have more precise control, oh boy...🔥
😃
I'm making songs with ai since february and the steps forward every month are unbeliveble. With this new features now is really starting to get fun. The only thing left is to have the sinth directly into Udio to control the pattern during the song.
that would be nice!
Need stems for each track. That will be game changer.
We need basic income now
yes please
The duet idea is great, can you get it to harmonise by using meta tags [harmonise 1 semitone] at said lyric
If not I'm sure that feature is coming soon. This is just in its infancy. Eventually you'll have a mind uplink where you can just imagine a tune in your head and it will use that. Not joking.
@@nickmcconnell1291 I’m really impressed with sonic in painting, so the program understands where location and phrasing is. I guess it would sync up with softwares that let you sing the phrase to trigger midi instruments. So long as AI doesn’t give people amnesia about thinking . Good old Telepathy outshines streaming. 🤪
@@pongtrometer 🤔👍
Great sharing and happy hearing your beautiful singing
lol thanks
Will it be simple as taxing all AI based revenue income to the hilt you will just lose money if you use it?
This what I wanted all the time. Generate me a song from my melodic input 😊
have fun!
Tuen on MANUAL MODE in your prompt bar at Udio, that is 100% more control.
idk if this is important. but it was very frightening because somehow my preplexity ai completely denied or they don't wany any AGI to live in this world when i asked it about AGI wether or not it should be arrive
What is the copyright legality of using the music output for any personal or professional project?
Does udio scan the output for copyright infringement?
Its Amazing and thanks to you your the master of Udio... i am from Ethiopia i wish Udio has Ethiopian instruments and can play our language atlist the Amharic one...
Wish you could overlay lyrics from the start of the upload track rather than just an extend section back or forth. Also i've not manage to be able to custom add lyrics for uploaded tracks, only seems to be auto generated lyrics add to an upload. I've tried numerous times with no success, just totally ignore the custom lyrics. Anyone else having this issue? Or is just me. Btw, i've got over 80 tracks on my channel that have been generated with Udio.
Great video and info. Both Udio and Suno still need some serious work on ducking the high strings behind the vocals when they build up and get louder. I get that issue in House music. I fix it in my DAW but it would be cleaner if the first production took that into consideration. It's new and will get there I'm sure. As a music composer and producer, I look at AI as an Instrument. Just needs to be learnt how to play it. Thanks for the video.
Thanks & good point. I'd love a DAW integration where the AI can generate/edit individual tracks, and then we can do the mixing/mastering
@@theAIsearch 100%. I think it's only a short matter of time before we start to see this. I can't wait and it will be a great help for every producer.
That nokia ringtone extended by AI actually sounded great!
agree!
What is the music software that you use?
cakewalk www.cakewalk.com/
Beautiful
Thank you!
whene we can use gpt 4o-?
you mean the new voice?
@@theAIsearch
yes, talking real time, like you show in demo videos
@@topremix0 it's not out yet
The song created from the singing clip shows an improvement of ai music composition. This is great except that its not free.
I think I agree with Rick Beato's son. (See the recent video on his channel.) I find it impossible to listen to this stuff for longer than about 5 seconds because it's like sandpaper on my brain. Hard to articulate why AI music is so instantly recognizable, but it's squarely in the uncanny valley.
A dream for content creators looking for free or cheap music but a nightmare for musicians
I am a musician , and believe me it is NOT a nightmare LOL
I still have to compete with other musicians and an unfair music industry., AI or not
If I get lucky and create something awesome it will kick ass on any AI in my genre.
Or at the very least compete with it and make me proud.
what system did you use to make the Riff again?
cakewalk is the DAW www.cakewalk.com/, vital is the synth vital.audio/
@@theAIsearch thanks.
Yea dude... This is absolutely insane..
It can add vocals to riffs but can it dd iffs to vocals? hmmmm
Impressive,but also whats the point of using this?
You have a good voice. Keep practicing
Around 17:00 , why does the song begin with the lyrics, and not begin with your original input sound?
I would expect it to begin there, since you cropped out the middle section and replaced it with lyrics as a “re-roll”.
i cropped out the beginning (with just my riff). see 15:50
17:40 lmao that moment was amazing. its incredible how much you can do with this.
could you share the full track? i liked it.
here you go: www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc
@@theAIsearch thank you so much. nice track !
Getting closer to Star Trek. Soon you'll able to use a voice command to tell AI to make music on demand, and probably a video to go with it too.
talk about hyper personalized content
Can you make the song bass & lights available to view on Udio ❤
here you go www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc
unbelievable - great stuff!
thanks
I always wanted to hear what came after that Windows XP start up sound... LOL
lol
you sing quite well !
🤯
If possible, could you post a video with just the EDM song?
here's the udio link if that works: www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc
@@theAIsearch Thank u! This is perfect!
That feature is amazing, hopefully we get more tools like this to assist in music production instead of just content generation, gonna try the paid plan and give it a shot
still a lot of hit & miss so far. i'm hoping for a DAW integrate and separate track generation
so powerful :>
yes!
Suno made upload feature free for all users today :)))
Wait until you hear my new AI song based on my flatulence samples reaching the billboard charts
You could upload your own crappy recording of your song with imperfect vocals, and get back a pitch corrected professional sounding recording of your song without paying a singer and recording studio or spending time with a home d a w. It's a time and stress saver
If anything it could be a creativity enhancer! No need to be an engineer and spend time fussing over a mix. No need to sing like Sebastian Bach. I'm excited about this. Because of my health I haven't trained my voice in years
Is your voice over from ElevenLabs?
nope this is my real voice
It would be cool if it could not only continue the uploaded piece but also "rewrite" it, meaning create an arrangement based on a given prompt while preserving the melody. This will likely be possible in the near future as well.
it can kinda do that if you use [riff repeats] or [verse repeats] but it's hit or miss
Hey brother, a new tts model has just came out, it is called "seed tts". It is actually a family of high quality versatile speech generation model. It has some other features too! I strongly want you to make a detail video on it, from showing it's use cases to the actual installation process! It will be very helpful for me! Thank you
thanks for sharing! ill check it out asap
Very impressive and slightly terrifying at the same time - like much of AI.
starry embrace sounds like a steven universe soo cool
i wrote an amazing hindi song...and i coudnt find a way to coninue this...this is helpful
Sonauto also has this function for free. But not as good. It has not copyright issues though...
ok so new you can prompt music genereation AI algo with music - makes sense
Great job on this video! UDIO's new Upload Feature is crazy good. I did a step by step tutorial video on my channel showing how you can take an 8 year old demo of an actual song written by Nashville songwriters and create a newer modern day version of the same song using Udio's upload feature. Technology is just getting better and better quicker than expected which is scary.
thanks for sharing!
this guy’s bold for making an udio video after that thumbnail a month ago. btw i used the new feature to make a song called Rejuvenation on my channel if anyone’s interested
Uploading original work would be in actuality swallowed up into Udio AIi same with Suno ect. Meaning, original work will be used to further strengthen and train their AI. Also, it's in their licensing agreement that you agree to when you join any site that any and all music you generate can be used to redistribute/sell to third party's platforms like TikTok. Which will undoubtedly happen.
If it's free you're the product
Great song 22:07, right? Much better than the other stuff. (Note i thought it was “beautiful lies” - that might be a great twist to incorporate somewhere.
it does sound like 'lies' now that you mention it. what a twist!
Udio is clearly trained with copyrighted material and without consent of the authors. If you ask it for very specific genres you'll recognize the voices of real singers and the sound signatures of important bands. The technology is very interesting but as a product it's problematic in lots of ways.
@eardrumfilms how did you learn to play? did you 'train' by listening to 'important' bands? did you ask the copyright owners of the music for consent?
But it's great to use it as "idea generator", then record your own vocals and instrumentals based on that.
It's just not impressive. A prediction engine isn't going to produce anything that hasn't been "heard" before. AI is glorified autocomplete in its current form, and won't "replace" anything unless new architectural innovations allow for more sophisticated inference.
@@Dom-zy1qy Yes I agree , you should be using session players with families to feed , not robots .But I'm sure my statement won't stop the Robot frenzy.
@@user-rr3by4bp2q Yes, that is the core of it, we are all remixing what we have heard and learned in the past. The question is - do we want tech to do it all? What will we do ourselves in the future? Nothing? Is this the beginning of the Matrix?
I always have said this, Udio Neural Network is simply Magic. Is Humankind Patrimony 😂
If they encapsulate the artificial voices to make them perform on other styles and songs and make them to speak connected to LLMs and Talking Avatars then people will be capable to create Artificial Artists.
Finally, somebody posted the full Nokia theme!
😃
Make a video for morph studio free ai
Luna is free DAW too
It’s lit!!
😃
now I'm wondering if Aaliyah wasn't an AI-generated singer from the future
Informative, can you make a 3D animation AI review?
thanks. ill see what i can find (though im not an expert in 3d animation)
@@theAIsearch
THX SO MUCH ♥️💖
This is freakin INSANE!!! Music Producers are gonna use this by "INCORPORATING" it into their already completed works! This is gonna invite some serious experimentation for advance Music Producers and Composers! But as I always say.....making STEMS from it is a must!!
agree. i'd like to see a DAW integration for creating individual stems
Hey, I didn’t get your name! But man, this is best UA-cam video I have seen in a longtime! Informative, instructive and fun! Great Job, I am downloading the chatLLM and signing up for udio! Hope you have an affiliate link, if you don’t get one quick! Again, thanks. BTW, I am not a musician or producer, just luv music! Let’s see if I can knock one out of the park!
how is this diffrent from suno?
they are very similar. i find udio's vocals are better, suno has more consistent melodies
AWESOME!
Thanks!
I need that windows theme
found it! www.udio.com/songs/8ACm7odbDDyuzEHyfJDexC
29:47 !!!! Just wooow
😃
The real fun begins when you feed one AI tune into the other. Suno into Udio and visa versa.
good idea