Master Udio with THIS - Make Better Songs Instantly
HTML-код
- Опубликовано: 30 сен 2024
- The full Udio guide. Udio tips, tricks, and prompt hacks to make better, consistent songs. How to repeat melodies, control sections, add effects & more!
#udio #udiomusic #suno #sunoai #ai #aimusic #aimusicgenerator #agi #singularity
Thanks to our sponsor Synthflow.
Synthflow allows you to create AI voice assistants capable of automating outbound calls, answering inbound calls, and scheduling appointments 24/7.
bit.ly/synthflow
If you enjoyed this video, you'll also like the sequel where I go over new features:
• This AI is a dream for...
Newsletter: aisearch.subst...
Find AI tools & jobs: ai-search.io/
Here's my equipment, in case you're wondering:
GPU: RTX 4080 amzn.to/3OCOJ8e
Mic: Shure SM7B amzn.to/3DErjt1
Secondary mic: Maono PD400x amzn.to/3Klhwvu
Audio interface: Scarlett Solo amzn.to/3qELMeu
CPU: i9 11900K amzn.to/3KmYs0b
Mouse: Logi G502 amzn.to/44e7KCF
If you found this helpful, consider supporting me here. Hopefully I can turn this from a side-hustle into a full-time thing!
ko-fi.com/aise...
If you enjoyed this video, you'll also like the sequel where I go over new features:
ruclips.net/video/X7DnlE-c2eI/видео.html
More updates by Udio:
Hi everyone,
We've rolled out a few cool updates to the platform:
Track length limit is now 15 minutes. Enjoy generating prog epics and extended trance mixes.
Context length for extensions has increased from 30 seconds to 2 minutes. This means Udio can "see" more of your track when generating an extension. In practice it means you can reference and repeat audio from further away in the track. Consistent choruses are now easier to achieve than ever.
You can now switch between the flat and the tree view in "My Creations". In the tree view, children tracks (e.g., extensions) are displayed underneath parent tracks.
Please keep the feedback coming, and let us know if you have any issues with these features in particular.
🎉 let's get the party started 🎉 amazing news 😮
Thanks so much for sharing this uldate news. I didn't know about it!
k just read this, i am so hard right now.
(also these are called parentheses)
I can't extend more than the 30 seconds at a time. The cropping is working. I am wondering if all of these features are rolled out or these things are still to come.🤔
It would be great to get a video explaining the cropping. I managed to delete all the work I had done whilst trying to t extend. 😥But it looks a great feature
For those looking for the meta tags taught in this video:
[Verse x]
[Chorus]
[Chorus repeats]
[Verse repeats] - Repeats the pattern more or less of the previous verse melody
[Interlude]
[Instrumental]
[ Solo]
[Voiceover] - Speech to text
[Spoken] - As if said by someone
[Laughter] - Like crowd laughter
[drop]
[buildup] [build-up]
[applause]
Hi, Thank you for that nice video. Here is roughly 30 tags or so you can try
[Verse]
[Chorus]
[Verse]
[Chorus-CHOIR ONLY]
[Chorus][CHOIR ONLY] Can sometimes work better if broken up
Add descriptive Style Words to metatags to guide how the lyrics should be sung.
[Sad Verse]
[Happy Chorus]
[Female Vocal Verse]
[Choir Chorus]
Use musical terms to influence the genre.
[Rapped Verse]
[Powerpop Chorus]
You can also also try to put it all together.
[Verse][Powerful Male & Female Vocals][Powerfull Choir]
or test to write it i different ways
[Verse, Powerful Male vocal, Female Vocal, female only Choir]
[Verse-Powerful Male vocal, Heavy guitar riffs and Drums, Female only Choir]
Had some great results with where first two lines can be Female vocal and the last 2 lines a male vocal in a verse for example.
Choir will sing words withing ( ) sings at end at most times then.
Works really good with symphonic or musical style songs.
Tips to test it with:
Symphonic Metal
Futuristic Opera
Broadway Musical
Pre-chorus and Bridge
[Pre-Chorus]
[Bridge]
[Shout]
[Whimsical]
[Melancholy]
Song Structure Tags
[Intro]
[Short Instrumental Intro] - you can try pretty much any instrument piano, guitar etc
[Talking Intro]
[Hook]
[Break]
[Percussion Break]
[Interlude]
[melodic interlude]
[Intro - Scratching record sound]
[MODULATION BUILDUP]
[CHOIR ONLY]
Outros & Endings
[Outro]
[Refrain]
[Big Finish]
[End]
[Fade Out]
[Fade to End]
[FULL EPIC CLIMAX]
[EMOTIONAL ENDING]
Instrumental Tags
[Break]
[Instrumental Interlude]
[Melodic Bass]
[Percussion Break]
[Syncopated Bass]
[Fingerstyle Guitar Solo]
[Build]
[Bass Drop]
Singing Style
Enter in the style description and as lyric prompts:
Gregorian chant
Melismatic
Narration
Spoken Word
Sprechgesang
Emotional
Sultry
Resonant
Ethereal
Lounge Singer
Vocaloid
Metatags
Style vocals metatags
[Female Narrator]
[Diva Solo]
[Gospel Choir]
[Primal Scream]
[Rap Verse]
Hope you all have a great day!
Much appreciated. 🍹
Awesome! Thank you!! 🔥🔥🔥
@@mannyj8190 sadly my hdd died and i had 3 new lists. One with +2100 song styles. Been sitting here now +16 hours trying to get the damn disk to work. Learned a lot new stuff and recreated all from EFI to volumes and disks. Think om close now but still getting blue screen but managed to get win to detect it again and running chdsk to hopefully do some fixes.
Might have a copy in my mail and can see if i manage to upload it online. Will get back and update if so. 😁
you are my savior! thanks !
@@ZenithZerksadly Windows broke my computer in the last win update but should be able to extract the most important files. Have now about 2100+ song styles and all sort of prompts to use within the song. If i manage to recovery the files i will upload them online and update here. Ah well back to try and re install win.... 😬
⌛
0:28: 🎵 Enhance music generation in Udio by utilizing hidden features and customizing lyrics for better results.
4:17: 🎵 Techniques for enhancing song structure and melody coherence in music composition.
8:59: 🎵 Enhance songwriting with consistent syllable count, instrumental solos, and editing out unwanted lyrics.
14:00: ⚙️ Enhance songwriting by using curved brackets to make singers echo specific words.
18:31: ⚙️ Enhancing song lyrics with humor and spoken word in a digital audio editing tool.
23:21: 🎶 Enhance music creation with Udio by exploring genres like country for realistic vocals and clean mixing.
27:40: ⚙️ Enhance song lyrics pronunciation in AI-generated music using acronyms in Tech.
Okay NGL this is pretty much the definitive video to take advantage of Udio. Focus watched till the end. Thx.
Thanks!
Focused watch is getting harder these days lol
Udio is just copyright infringement, just blatantly extracts other artists voices to make songs.
rap producers are probably going nuts with stuff like this. Render 60s or 70s type music and sample away with no lawyers coming after you.
My biggest problem with udio is the moderation restrictions. Let me write mature lyrics.
What precious advice you gave me with this video, before to know about [Chorus repeats] I was always write in the prompt "the chorus must always have the same sound and rhythmic " and always worked
First radio play of a Udio song was last night on Oystermouth Radio at just after 9:15pm. The show should be repeated tonight at the same time.
When will we be able to control the chord progression, time signature and tempo? This shouldnt be too hard for an AI to provide, given what it is capable of now.
What about advanced features like seed and prompt strength? I haven't had any luck with udio. The songs sound horrible.
Your tips have been incredibly helpful, and I truly appreciate the time and effort you've invested in sharing them. However, one major challenge I encounter is the inability to edit lyrics within the initial 33 seconds while maintaining the same musical accompaniment. This limitation forces me to either create a new 33-second composition or remix, inevitably resulting in a different song or melody altogether.
tech still has it's limitations
They recently increased the context length from 33 seconds to two minutes. That's why consistency works now. Greetz
Some really helpful stuff in here! In the end it's all taste, but personally I think the stuff it generates when you do simple prompts like pop/country/folk/etc is really generic and boring. I mean, I could listen to any playlist on Spotify and get the same cookie-cutter music. You get much more interesting stuff if you give it one or two genres and then add descriptors. Words like Ethereal, Emotive, Pensive, Melancholy, Joyous, Mechanical, Innovative, Strange, Uptempo, Downtempo, etc will give you much more interesting music. It's hard to really control where it's going using many words, but honestly I think that's the fun of it.
Thanks for sharing these keywords. I'll try them out!
You definitely need Timestamps in your video
I'd love a similar rundown on using the prompt field. I'm certain there must be more control to be had there. Great insights so far, thanks!
Did you ever find that info? I can't seem to get udio to do male then have female (or vice versa) then a duet. If I had that I could work with it.
How to do the same voice in other track???
Am I experience watching videos and using the free version you need to pay for either the $10 a month one or the other one in order to be able to go from a song that’s already existing or in your case of Voice that you like
Can UDIO make acapella vocals for my own remixing? I just tried using udio today for the first time and it created great songs in funk and disco pop styles. I would like to get the vocals but the online stem splitters I tried don't get the voices out cleanly.
Great video. I've learned a lot, but I have a question. Can you force song structure for instrumental songs (no lyrics)? For example, I'm trying to create an instrumental and I want to maintain the themes from previous sections but the AI just wants to take the song a completely different direction. Basically every new extension sounds like a different song. In the video, you were able to maintain the the chorus and verse melodies, but how do you do that with an instrumental?
Exactly what I wanted to ask. But would it not work if we write in the lyrics field something like [theme 1] and then [theme 1 repeats] in extensions? Without any other words. Will the AI still add voice?
Pop blues sounds like 1960s Motown music nothing new.
i got great results with 60s soul, delta blues, new wave 1980s, 60s easy listening (in the style of burt bacharach) grunge, i do think its worth using "in the style of" because it adds new prompts to direct it towards that style.
Thanks
Same with contemporary, modern country. Hard rock and metal was kinda meh.
Actually the artist one you mention that it can't do, for me personally it still finds something close to the style of the artist even though it says it will replace it. It just finds something similar to the style. Also, sometimes it puts words in between in the song, which I can't remove, if I like the parts on the other side of the song, which sucks. I sometimes use manual mode because it works better for what I want. It kinda forces it to be in a certain style and minimise the risk of it using words I didn't write in the lyrics. That's at least my experience.
I hope they one day makes it possible to get a karaoke version of the song written, so I don't have to use a vocal remover, since that isn't as perfect as having the real instrumental of the song.
Sorry but you really give false advice regarding to "fits within 33 seconds" - This frame doesn´t matter, it is the 130 seconds context window which matters. When the chorus or any song-section is cut in the middle by the end of a generation, it will just seemlessly go on in the next generation, as long as you don´t change the prompt too drastically. Song structure doesn´t need to correlate with the framework of 33 (now 32) seconds, and it would be super annoying if it would. 130 seconds is the length you need to keep in mind.
This is a great class on using Udio, which i happen to use professionally now!
spotify will be flooded by ai music
They have been clearing a lot of stuff like that from their service.
@@ProckskiWhy? As long as people have the rights to it and it's good music...
Question is, how do they find out it's AI
@@pocketpheonix9573 Well, one giveaway would be if they add the words Suno or Udio to their song name/descriptions. I believe that's required in order to legally publish
I have already started a music company now working on it heheh
instead of BPM have you tried saying 'uptempo' or 'waltz rate' or something?
or perhaps something "allegro" . . .
I made a song with vocals and a LEGENDARY instrumental that was so good, but I liked the instrumental so much that I just wanted the instrumental, and take the guy's voice out (from a song I already created) Can I get the instrumental of an already created song that has lyrics?
Use a Karaoke tool.
this help me alot back to the shadows i return
enjoy!
Some Deathcab For Cutie vibes at 12 minutes there! 😂
This is a nice tutorial that answered a few questions that would have taken me a bit of time to discover on my own!
Nice job!!!
I think this service is really fantastic for creating some inspirational material to work with as a base or starting point without spending hours noodling around on a synth or guitar and basically just having the musical equivalent of "writer's block" 😋
very helpfull :) [chorus repeat] ... thanks :)
my pleasure!
I tried with things like [guitar solo] or similar with brackets, but I got only a voice that sings "guitar solo"... not what I asked for.
for Time signature, try typing in Mixed or Complex or Compound Time Signature. It Also understands Poly-Rhythmic. for rhythm add straight or swing.
Thanks for sharing!
Waltz also works
Great tips, thank you!
many good things here in this video, bravo! thx alot!
my finding: there is a prompt for scenery sounds, what i dont know how to use or what it could be.
e.g. i tried to make a rocksong inspired by the "Dune" movies. and what udio did, was adding Battlesounds at the ending of the song (Laserguns, screaming, explosions), without any input by me (just made an instrumental outro). actively trying things, like [sound of sandstorm] or something didnt work at all.
Thanks for sharing!
Title: Make Better Songs
Thumbnail: Taylor Swift
After seeing this video, I started using Udio and here are some songs I produced with it
my words:
www.youtube.com/@magnifico372/videos
ruclips.net/video/-nv3TM2zaVI/видео.html
Udio's words:
ruclips.net/video/XMLjVK03wOo/видео.html
ruclips.net/video/zjxLDsKvZEY/видео.html
with a poem by Baudelaire:
ruclips.net/video/CF3sPTgMSmM/видео.html
As a non-native English speaker I prefer Suno , because it has more capabilities when it comes to customization and using non English lyrics. I hope udio start to improve on this one!
same here , i prefer Suno
Hey, could you say more about what you mean? I'd love to know more about how Suno is better or what it offers
@@WaitingForTheHook for instance, you could put an Arabic lyric or Text in Suno and it ACTUALLY sings the lyrics very well, but in Udio it doesn’t work significantly for Languages besides English, another thing in Suno is that The Quality of the songs got improved ( at least in my experience ) , so yeah overall Suno is better than Udio in my opinion just cuz it works for me. U should try it.
This is a bit of a short coming for every model, Suno included, for your use case though this makes sense, Suno has been trained on many more langauges if im not mistaken, and Udio has been trained on mostly the popular ones, it does Brazilian okay, and Latin perfectly biut if thats not what your going for then eh
@@WaitingForTheHookSuno also generally sounds better, but this is coming from a music snob. I notice less errors with following the prompt and overall less randomness. Sometimes uduo takes me 7-10 tries just to get the genre mix right where Suno hits it the first time
Udio creates music with a real singer's voice, in my language I asked to create music and to my surprise it was the voice of a famous singer in my country
Who was the singer and do you have the song to compare?
May happen when the data base for that country is small. If the singer is too "famous" and did a relevant % music, machine "learns" that a % of the times a singer shold be have such a voice. Is up to you NOT publish that song because it would be a violation. (is the same if I asked to sing a song of mine like "sinatra" to a human imitator, then I can use for my personal listening but is not right to commercialize it!)
this is because they trained their models on copyrighted material. If you don't want any legal issues stay away from it
It's not music lmaaao.
@@ruxboxyeah I'd never publish anything under my name, I only generated prompt based ripoff songs to see how far I can make it rip others off.
awesome, best tutorial on Udio i have seen, thanks 💯🤙
Subscribed 👈🤙
Thanks for the sub!
This tool is absolutely incredible. So fun!!! I've never had a good voice and my days touring with bands is over. This allows me to still "make " music, but better than I ever have before.
enjoy!
What if I generate music while I am not subscribed? Who owns it?
What if I generate music while I am not subscribed, and after 5 minutes I subscribed? Who owns it?
What if I generate music while I subscribed? Who owns it?
What if I generate music while I am subscribed and forget to pay my subscription for the next month?? Who owns it?
Do I need to give credits even when I'm in Paid plan?
What if I generate music while I am not subscribed, but I don't want to give credits, but I didn't know that I need to give credits, then later I noticed that I realized if you're a free user, you need to give credits, can I subscribe so that I can have the right not to give credits?
With the chorus you don't have to use repeat. It will automatically try to repeat it. But I haven't tried to add repeat to a verse I hope that can help. Thank you!
Also for those watching, Udio has since been updated since this video was made. It now doesn't matter if a clip of a song ends in something you don't want...because you can crop it out and extend from the part that you choose.
Also when generating a genre prompt, you can actually just type in a description of the sound you want. Like instead of using tags like Country, Rock, Classical music, you could write "country rock song with a classical influence" similar to how AI image prompts work. I have gotten good results with this method.
I want to address you saying about those genres that don't work out well in Udio...I will say it is less consistent than Suno, but when you get a good seed, it will be far better sounding than Suno because it's in stereo and vocals have more breath. Also like I said, describing the song style/sound you want can sometimes be more effective than putting in tags.
Something that's not touched on in this video I want to address...with Udio, if the lyrics that you write don't rhyme, it has a very hard time making it sing. This is one of the things Suno still does far better than Udio. So if you have a song that doesn't rhyme, you need to be well under the character limit, at least 100 to 150 characters under. So if you want Udio to sing lyrics that don't rhyme, it should be no more than 200 characters per section. You may still have to try a few times to get a seed that will sing as most times it will just be spoken words mixed with instrumentals.
Thanks for the update, Georgina.Jungle!
As long as you use these silly clickbait "do something with THIS" titles, I will never subscribe to your channel. You're not a buzzfeed listicle farm. You have good content. Just put in the title what you are showing. This stupid "THIS" trend needs to die.
if u want an AI girl that is only talking with you and not 1000+ people simply use a local LLM and its all yours
😃
🎯 Key points for quick navigation:
00:00:14 *🧩 Common user complaints about udio include lack of customization, inconsistency, and inability to repeat melodies or choruses.*
00:00:28 *🛠️ The video will provide tips to improve udio song generations significantly.*
00:01:18 *✨ Customizing lyrics manually ensures control over their structure and how they are sung.*
00:01:48 *🎼 You can import lyrics into udio, breaking them into sections like verses and choruses for better song structure.*
03:57 *🎶 The 'verse repeats' and 'chorus repeats' meta tags help maintain the same melody across different sections.*
05:57 *🔄 Repeating the exact melody for verses and choruses is possible, contradicting claims that udio lacks this feature.*
06:57 *📝 Ensure similar syllable counts in repeating lyrics for easier melody replication by the AI.*
09:05 *🎸 You can add instrumental solos (e.g., guitar or violin) between lyrics, but reduce lyric count to fit within 33 seconds.*
11:12 *🚫 Prevent unwanted vocal ad-libs at the end of a track by forcing it to end with an instrumental.*
12:40 *🎤 Use meta tags like 'interlude' or 'instrumental break' for smooth transitions between sections.*
13:40 *📢 Customize harmonies or echoes in lyrics using curved brackets to specify which words to echo.*
17:17 *🎙️ udio can also generate speech instead of songs by using 'voice over' or 'spoken' meta tags.*
20:21 *🗣️ You can simulate stand-up comedy by inserting 'laughter' tags, which generate crowd laughing sounds.*
21:49 *🌐 It’s best to keep the prompt simple; more complex prompts can confuse the AI and result in poor generations.*
22:11 *⛔ Specifying tempo (BPM), key, or time signature in prompts doesn’t work reliably.*
24:18 *🎻 Genres like Country and Bluegrass yield highly realistic and well-mixed results in udio.*
25:25 *🎭 Broadway musical style also performs excellently, providing dynamic and high-quality vocal outputs.*
26:08 *🎶 Suno may provide better melodies and consistency in certain genres compared to Udio.*
26:36 *🚫 Udio cannot generate music in the style of specific famous artists.*
27:04 *❌ Udio replaces artist names with keywords, making it unable to mimic specific artist styles.*
27:32 *🔤 To control pronunciation in Udio, use capital letters or separate letters with hyphens.*
28:13 *👍 Udio can pronounce acronyms correctly if formatted properly.*
29:09 *🛠️ Adding hyphens between letters ensures accurate pronunciation of acronyms in Udio.*
30:18 *🕐 Hyphens can also be used to create pauses in lyrics for better control of timing.*
30:44 *🔡 For hard-to-pronounce words, spell them phonetically in Udio.*
Made with HARPA AI
Can I upload a music to the website? then extend it?
I just deleted my music by mistake but luckily I downloaded it before deleting it 😌
I just want to upload it again 😅
Nice tips, but hearing these AI lyrics makes me want to puke. :P I know this is just for an example, but 90% of people are too lazy to even write their own lyrics and it has this specific style.
Does any of you guys know if udio could use my own guitar track to build a song? I mean that it will use my original guitars and not only will generate a song inspired by my guitars. If somebody knew how to make that happen and explain it to me i would definitely duck your sick!
00:02 Master Udio ermöglicht die sofortige Verbesserung von Musik
01:51 Importing and organizing song sections in Udio
06:04 Refining and repeating melodies in digital music
08:07 Creating emotional depth in music through programmed beauty.
12:26 Extend a track seamlessly with instrumental breaks and interludes
14:33 AI technology allows for the creation of human-like AI voice assistants.
18:11 Transform song into standup comedy with added spoken elements
20:05 Deactivate manual mode to improve output quality
23:25 Musik wird lebendig mit klaren Vocals und sauberem Mixing.
25:29 The AI-girl with a soul captured in coded art.
29:11 Optimizing letter spacing and pronunciation with CSS and JavaScript
30:59 Experiment with these tips to improve your songs
Thanks for a great video. I hope you'll forgive these two pedantic points.
"curly brackets" are called "parentheses".
And that "country" song that you said udio did very well was definitely bluegrass.
"after listening to BOTH" made me giggle, because my workflow is very different: I am usually going through at least eight variants when generating a new section, sometimes even 16-20, when a song comes to an important section with high novelty factor. and then I often just take about 10 seconds and then cut, instead of 32 and also do a lot of inpaintings, especially in the instrumental sections or instrumental songs, That is a slower progress, but especially when following an approach of creating complex yet consistent music, the results benefit from being picky, as they become less generic and more unique and aligned with your personal taste and vibe.
By the way, you can also do the prechorus and then the chorus in the next section. The 32second frame is irrelevant to how you structure your song, the context window should be your reference here, and this is 130seconds - or shorter, if you change the settings. As long as verse-prechorus-chorus fits into 130 seconds, you´re safe. The 32second frame doesn´t matter regarding to song-structure, I think that´s important to understand!
The tip with adding "repeats" is interesting, though, gonna try that out, thanks. :)
This is not music production.
Nah. It’s better
I disagree that EDM doesn't work well. I've made some great EDM tracks. I think the key is to add emotional cues to the tag like: dark, haunting, ethereal, celestial, alien etc Add a time period like early 2000's, retro 80's. Add instruments like Roland TR-808, Korg synths. And then other details you want to influence it like: repetitive, glitchy, industrial, mechanical, melodic, pumping bass, fast tempo, slow tempo etc. Listen to a song you love and write down everything you hear and feel and use that as the basis for the prompt. If you just write "pop" then you get very generic results.
Can't Udio make the lyrics? Why bother importing from Chat GPT?
You are the Boss Man!!
Chatgpt 4 + Udio AI + Tensor art AI
ruclips.net/video/DXpihH0IHIA/видео.html
Nice for fun but safe to expect some possible lawsuits in the future. Please do read T&C's you also have to contractually credit Udio (but ironically not the artists it's trained off of) but have fun.
I've got 45+ songs under my belt so far and I can point out some mistakes you made: 1. you are giving the AI too many lyrics to sing, no human could sing that, so its unconvincing from that perspective. Also if you make the AI sing lyrics that are awkward to sing, you can try it before the AI tries to do it, if you can't sing them, don't bother giving them to the AI. One of the things you will noticed with UDIO is if the lyrics are unreasonable it will actually ignore the lyrics or replace them with gibberish.. So you want to be sure your lyrics give the AI enough room to sing them. In this case it works kinda fine cause you picked a genre that permits a lot of lyrics, the mid-60s Motown sound. I just finished off a 7 minute song sung by what is certainly Glen Frey's voice.. I'm uploading it now.. If you listen to it, I feed him a lot of lyrics, but I was pretty lax with the ways he could sing that, so some are normal, some are stretched out, some are quiet, etc.. I really explore the space so to say.. I also noticed at the beginning of the song the AI vocalized your Intro tag. That may be that the intro tag was not properly written, it probably used a curly bracket than a square bracket. UDIO should ignore the square brackets.. the Interlude one I learned about this morning when I made a Church Hymn for capitalists at the First Church of Ayn Rand. And I needed to glue the chorus to the next verse and it was ending the song after the chorus. I put an interlude after the chorus and it filled in the space between the chorus and next verse. One of the difficult things with UDIO is when just suddenly decides the song is over and ends it early.. There is now a commercial version of UDIO, I have the pro membership, and it has so many credits you'd have to be making 3 songs a day for a month to use them all. The $10 membership is suitable, its 1200 credits I seem to recall and thats enough for about a couple of CD's depending on how satisfied you are with the outcome.. The more disatisfied the more generations.
Also those with free membership can only generate two parts at once (normal), those with the standard membership, can have 4 segments rendered at once. And pro members can render 8 segments at once. And since I have pro I have been pushing it.. However you have to wait until it finishes off all the segments before starting about 8, cause it will give you errors and you will be left with zombie segments, and usually have to delete them. There is a tree mode that expedites creation, but it only can show you a tree with less than 100 segments cause the tree is just a view atop the list view and the 100 segment limit prevents the complete rendering of some trees. If youa re trying to get a particular musician, ask your lyric maker, mine is gemini (I used to use chatgpt plus, but its performance has gotten worse, and I had to leave it for something more dependable and gemini is, its also much easier to copy text from and share discussions, I think chatgpt requires people to be logged in to see the chat link.. Anyhow, I'm satisfied with Gemini.. It doesn't make pictures, it can, but its just an added distraction, if I want to make pictures I will go to huggingface.co and use one of the many demos or the new story diffusion which is supposed to outperform SORA.
Anyhow, Gemini makes my prompts for the song style, and my lyrics.. And its a heck of a lot more accurate than just picking genres directly from the creation window. I was messing around with stevie nicks andd crosby stills and nash earlier, and CSN would do the creat harmonies so I gave up and extended a seed they I discovered was Glen Frey.. It just happens.
If you get the guy on the horse at the bottom of tthis playlist on UDIO, thats the Glen Frey one I just did.. HE is the guy who sings smuglers blues and peaceful easy feeling. This is my dopplegangers playlist, all these songs are representative of real musicians, you just have to listen to determine who was emulated.. www.udio.com/playlists/5w6ykXN5pBeyUpyg8J6gJm
I just started messing around with this Udio Beta. Some hits and some misses as there is a large learning curve. Worst thing is that I can love how the song is sung and played, but hate where the lyrics are gibberish or it cuts off or allows huge dead space. Sad thing is, there is no way to recover the styling, despite the reused prompts.
It's fun, however should not be mistaken for proper music nor production,. as it sounds like shit, from a sonic perspective. On top of the fact its learning from music already out there, so the moral of this is not sitting well with people.
I am really amazed by these tool. I am creating music in Italian and Spanish. Result is cool!
When I stem my vocals from Udio songs and run them through Kits to change to another singer, they always cause the conversion to glitch out and randomly go up and down in pitch like bad autotune. The settings on Kits do not matter, always the same result. Probably because of certain fx and "fuzziness" from the AI generations. Is there a way to EQ or filter the vox track first to smooth it out in a way that will then be accurately interpreted by Kits for a good and useable conversion?
What is Kits?
This is hands down the most helpful Udio tutorial I've seen. Many thanks!!
You're very welcome!
great video, very informative, but your (speech pattern?) makes it hard to watch. i dont know if youre editing out gaps between words or literally stopping all sound and putting a small break between your words a majority of the time, or if maybe youre using some audio compression that is making it go dead silent between some words... but it makes it really hard to keep watching. doesnt feel like your words flow naturally together a good portion of the time. if you could tweak that one little thing i feel like the overall watchability would greatly improve.
great video though with great info!! thank you so much.
Even though I forced udio not to add new words to the lyrics using [Instrumental break] it kept adding its own lyrics which is a cacophony of sounds that sound like words but incomprehensible. My song is in Russian maybe because of that?
Mmm, cool tech, sounds pro but awful songwriting! Didn’t hear a single chorus!
In real songwriting, choruses are all about crafting expectation, tension and release. Ai needs to be able to read between the lines of subjective human expectation to be able write a decent song.
In simple terms, this songwriting is far too objective.
why not just have Udio write the lyrics in app? they are just making an api call to gpt4?
same thing with extra steps
>write your own lyrics
just sit and write.
now work with the model to refine your lyrics
zoom in on a line and iterate on the lyric
now use this for your gen music.
the technology is far more powerful when it’s sourced from your inner cosmos
at least journal a powerful memory or a suite of ideas and workshop this with claude or gptomni.
keep on tagging
you want a generic song? pick like two or three
go deep.
like:
dream-pop, wistful, female vocalist, longing, heartbreak, warm
this is when you find your gen music hits a note and the magic happens
Udio may have the resources to take on the major labels in a copyright case, .11:56 . . .but could they win A-ha copyright case if they Take On Me.
I have three spotify artist accounts fake names obviously I'd rather remain anonymous in that aspect, so I use this software And I am up into the hundreds of thousands of streams On my accounts and All done with a few prompts. All the artwork was done with a few prompts, with very little effort and I have released 5 albums now in only a few months. I've lost count of the amount of playlists the songs have been put on But there's hundreds.
All done with a few prompts.
There's a tip to do both auto generation and lyrics simultaneously.
Use prompts like: pop rock, add Lyrics [Verse], [Chorus repeats], [Interlude] etc.
Put it in the prompt box itself.
Then keep auto generation on.
Lyrics Strength high
I know it sounds weird to put a full on lyrics inside the prompts, but I managed to get it to say the prompts itself with the lyrics.
Of course, it is not always work the way you want, but I personally found good results
Great video man! Thank you for making it!! Kudos!! ⭐️⭐️⭐️⭐️⭐️
My pleasure!
Hey Shiey, is it you? The guy talking in this video sounds just like Shiey: www.youtube.com/@shiey
He's also into music, so I wonder 🙂
I've been playing around with Udio for awhile. Is it possible to do some enhanced prompt generation?
Dude, you didn't read the docs well: turn on manual mode and Udio will save "in the style of..." However, depends.
Go learn music. Or else be a fool using a program to make a copy of a copy of something. Music is supposed to free us, to teach us, and to be a vehicle of expression - don’t let a computer express your feelings
It has very wide palette of styles and does convincing arrangements but it has sub-par skills when it comes to writing strong melodies.
This video is exactly what I've been looking for. Well done.
Thanks
Anyone currently having the issue when deleting tracks? It seems like the delete option isn’t available
Is there a list or repository of meta tags you can provide?
Udio makes way better lyrics than chatgpt by far, for me.
Le "repeats" n'est pas utile, dès lors qu'un premier couplet ou refrain est réalisé, les suivants seront sur la même base mélodique.
If you want to make music, go and learn how to do it. Instead of flooding the market with this AI nonsense
you guys are such amazing musicians, incredible talent! 😃
must be a joke for shure. Letting an AI generate the music does not make you a musician. If anything you are a moccery of a musician. Disgusting shit this is. DOnt even think of calling AI honks musicians if you dont want to insult every real musician here on the planet
Just kidding?
😂😂😂😂😂😂😂
let me know when ai can make a HIT. so far iv'e only heard generic stuff.
That song sounds a lot like Sugar Pie Honeybunch.
It's unbelievable that you use so many AI tools, cover the best of them, and even give tips as a power user
Thank you. Have subscribed for many months now, and haven't had the need to check out other AI-related channel since you cover all the newsworthy parts of AI news and ofc AI software (still using the vocal extractor and vocal trainer)
Thanks, I appreciate it!
A good lyrical song is all about those phonemes and syllables. 👍
Is there an option to have voices based on age? Like kid or child’s voice?
These sound like some pretty useful tips! Do all of these meta tags work in Suno too?
Bad advice, this is not how you use this. Simple styles = Simple Music.
You start complex and keep generating, an adapt it to a working prompt. This is why you can only get 4 style to work. Ill comment a song I have made below.
Song on this profile, I made this generating advanced style prompts through the magic prompt.
The style prompt for these originally was blink 182, xxxtentaction, emotional, same prompt on 3 of the 4 songs on this account. After Udio replaces those 3 artist with tags in random combination.
Eventually the AI gets a combination that works, this is what you are looking for and you copy and paste that series of tags. Keep generating until you get something closer. Then you can start remixing that. Get it even closer to what you want. This is how you get R'B and trap or a merger of Punk and Hip hop like I did on this random like juice world mini album. You don't want to avoid complex prompts you need to learn the AI and it understands.
good crap... it's good for bad musicians or playback artists
can someone make a short list of all metatags, and vocal style, and voice effect like
[intro]
[outro]
[verse]
[chorus]
[pre-chorus]
[Instrumental solo]
[Interlude]
[chorus repeat]
[verse repeat]
[laughter]
[downtempo]
[drop]
[Build-Up]
[standup] in prompt
[voiceover] in prompt
metatags : []
repetition in lyrics : (word to repeat)
pause in the lyrics : --- between words
prompt : voiceover only
thanx, you da man, although i typed out all his tags myself and when i saw your post i was Doh!!
so hopefully your post gets boosted to the top 🤙
You just did! Thank you 🙂
Great video. Is it possible to get Udio to create a song from my lyrics then, after has done so, take out the vocals so I could sing the lyrics myself?
the second i hear taylor swift its a dislike . good video but nup i dislike that a lot.
I dunno Suno V3 still looks better imho, plus Suno has max 2 minutes per song plus you can also extend it. I also tried it with lyrics with different languages and its works very well.
The Suno voice just sounds too generated to me. Suno is very consistent though, I'll give it that
How do I take a section of one mix and import it into another song?
11:50 Can I have the sample? It's nice, I want to use it.
CAN U MAKE THE SAME VIDEO BUT IN DIFERENTE LANGUAJES TOO MUSIC IN SPANISH OR JAPANESE ANIME MUSIC AND USE THE NEW UDIO UPDATES ANS ALSO USE TWO VOICES IN THE SAME MUSIC
As cool as this tech is, its really going to devalue music.