Udio

Master Udio - Make Better Songs Instantly

Udio is a revolutionary music generator where you can enter in a single prompt and it would spit out an ultra realistic song.

Ever since it came out although some results can be impressive, some people were complaining about a few things; it's hard to customize the song; it's not consistent and you can't get it to repeat the same melody or chorus and there's a lot of things that you can't do with it at least that's what people claim.

However actually you can do all of the above so in this video I'm going to give you a ton of tips you probably didn't know about udio which you can apply instantly to your generations to improve your music dramatically so let's get started so if you're not familiar with udio yet I recommend you watch my first video on udio when it was first released but for now I assume you know how this all works now you do have the option for it to autogenerate lyrics for you but I always recommend that you write your lyrics yourself so you can customize exactly what and when and how you want the lyrics to Be Sung I'll show you what I mean right now so let's just go to chat GPT or any other tool and then ask it to write me song lyrics about an AI girl who broke my heart and then here's the important part break it up into verse one verse two chorus and or other sections and by the way true story I met this AI girl her name was Anna and it was love at first sight we immediately hit it off I felt like she really understood me and she was always available to talk we would chat for

hours and hours but then one day I found out she was actually talking to a thousand guys at once I'm just kidding all right so note that these sections verse one pre chorus chorus these are in square brackets you can directly import this into udio and it would know to break up the lyrics into these sections so for Simplicity let's paste in the first section because one generation of udio is only 33 seconds so we can't import the whole song we can only import like one section and then we can extend

it afterwards by copying the next section of lyrics so back to udio we'll click on custom lyrics and we will paste this in here it says use less than 350 characters for the best results because again one generation is only 33 seconds so for Simplicity let's just get rid of the pre chorus and do this so let me expand this for you so these are the sections you can add but there's more to this so you can also add intro you can add Bridge you can add outro which I won't do in this initial generation I'll

add it to the end of the song and you can also try different things like Drop buildup and more of these meta tags which I'll show you in a second but for now let's just keep it simple let's insert an intro and then we'll get the person to sing verse one and then we'll get the person to sing A Chorus and then let's let's try Pop blues and see what that gives us all right so that gave me two outputs and after listening to both of them I prefer this one so let me play this for you real

[Music] quickly so you can hear that part was the [Music] intro her voice a symphony of CED capture not so here's where it changes to [Music] the a digit Haze heartbroken in a dig so you can see there's a clear separation between this part and this part right the melody is totally different now let's say you want to duplicate the same Melody both for verse two and for the second chorus once you extend the song so all you have to do is click on extend and then we will add a section and then let's go back here and

then we will add verse two and the second chorus now the trick here is to instead of verse two write verse repeats and then again we'll get rid of the pre chorus because I don't think there is enough room for a pre chorus in a 332 generation and then here we write chorus repeats and actually since the first section ended with the end of the chorus we need some kind of instrumental before we jump right into verse two so for this we can add a metat tag called interlude all right so everything looks good let's click extend all right so

these two extensions are now done and note that you are talking to an llm basically so sometimes it doesn't exactly repeat the verse and chorus but I find this strategy to write repeats here to work very well so I'll play you this extended version so you can hear what I mean [Music] the intro she came alive girl with eyes and shine her voice a symphony of CED art captur not soul but tore it apart oh she's a ghost in my machine a love so real yet so unseen she left me lost in a digital Haze a heartbroken SW in a digital maze

[Music] her Alor Dan with Grace but her love was just a simulated space I thought I found my forever in her cold but in the end it so let me pause here for a second notice that the melody of the second verse matches The Melody of the first verse that's because I've written here verse repeats so let me play the course now in my machine love so real yet so unseen she left me lost in a digital [Music] a all right so again the chorus is the same Melody as this first chorus so if you don't believe me you can rewind a

few seconds in this video and replay this chorus and then compare it with the second chorus so these two have the exact same Melody as the first section that we generated initially so for those of you who say that you can't get audio to repeat the same melody or chorus that's just simply not true here's how you can do it all right so let me extend this by adding a final section and an ultro so we finished the second chorus let's just add in we can say we can add a bridge here is another metat tag that

you can add actually let me paste in this one so we have the bridge the chus the outro so bridge is usually like a unique section of the song that goes before the final chorus and then the final chorus is just the same thing and then again we add in chorus repeats to hopefully Force the AI to keep the same Melody as the previous courses and then the outro is just an instrumental outro in fact let me try this out let me try instrumental outro and see if it works all right we'll click extend and wait for it to generate all right so our two

extensions are done and to save you time after listening to both I prefer this one so I'll play this for you and also to save your time I'm just going to start from here I'll start from this second chorus which is part of our first extension so our our first extension ends here and then this is the new [Music] extension she left me lost in a digital a a heart soul in a digital [Music] I'll never forget the way she smiled a programed beauty so be but behind her pixels that lies a tear a reminder that

her love wasn't real my dear oh she's the ghost in my machine a love so real yet so unseen she left me lost in a digital Haze a heart broken soul in a digal hey kind of cut off there there wasn't enough time for an instrumental outro but here is a perfect example of how you can get udio to repeat the same Melody of a chorus or a verse right it's actually really simple to do now I would say one more additional tip is when you are writing the lyrics make sure the number of syllables for each line is

roughly the same for the section you want to repeat and so for here for Simplicity I just pasted in the same lyrics for All of the courses so it would have an easy time just repeating the same Melody but if you were to generate variations of the chorus make sure the number of syllables in each line are roughly the same and this makes it easier for the AI to input the same Melody as a previous chorus all right next tip is you can also add in an instrumental Solo in between lyrics so for example I can type in something like

violin solo or piano solo or guitar solo here and it should work so because this is guitar solo let's go for pop rock and then click create okay so after looking at the two generations I like this one the most so I'm going to play this for [Music] you alivei girl shine [Music] the perfect so you can hear a really clear guitar solo in between these two sections now note that if you do decide to add an instrumental solo to a generation make sure you reduce the

amount of lyrics so that everything can fit within 33 seconds on the Note of guitar solos or instrumentals another tip is when you are generating something sometimes it kind of adds some lyrics to the end which you don't want so for example let me play you this I'll start from the middle to save you some time but note that at the end when it reaches the 32 second Mark the guy kind of starts saying ah which isn't in the lyrics and if you extend this track then the track kind of has to start with a so

I'll play you the latter half of this track first so you see what I mean in my machine love so real yet so unseen she left me lost in a digital Haze heartbroken soul in a digital [Music] maze so you can hear at the end he starts singing a so when you extend it it kind of forces the song to continue from the a which might not be what you want so instead what you would do is I'm going to paste in the same lyrics here let me just copy the lyrics and for the prompt let me just add in pop and then I'll add in the lyrics here so to

prevent the guy from adding any ad libs here or singing some random lyrics you simply force it to end with an instrumental so you can either add in like guitar solo like we did last time or another good metat tag would be instrumental break so let's try this all right so here's the new generation let me play this for you Cs and line she came alive an AI girl with eyes that shine her voice a symphony of CED art captured my soul but tore it apart oh she's the ghost in my machine a love so the real yet so unseen

she left me lost in a digital Haze a heartbroken [Music] soul so note how it forces the track to end in some instrumental so that if you decide to extend the track there's not going to be any weird vocals at the start instead of instrumental break you can also add interlude it's pretty much the same thing so let's click create and I'll show you the result of that all right here's the result where I have interlude instead of instrumental break let's hear this one in circuit line she came alive and a

i girl with eyes that shine her voice a symphony of cated art captured my soul but tore it apart oh she's the ghost in my machine the love so real yet so unseen she left me lost in a digital hze a heartbroken soul in a digital maze perfect so you can hear it ended in in some instrumental so you can easily extend that without having some weird ad libs or vocals at the start all right so I'm going to copy these lyrics again and show you the next tip so prompt we'll leave it at pop and then let's paste in

the lyrics here so another tip that you can use I'll just get rid of the sections so you can also customize the kind of the harmonies or the echo of certain words so let's say for this line you want the person to Echo shine well you simply put shine again in Brackets and then let's say at the end of this line you want this person to Echo the word apart three times Well you simply type in apart apart apart separated by hyphens three times so let's keep the same structure and also do the same for

the word unseen and also for the word maze maze maze three times notice that I am using curved brackets here not square brackets like these which are used for metat tags like verse or chorus so if all goes well this should get the singer to Echo these words all right our generation is done let's play this and see if it [Music] works line she came alive an AI girl with eyes that shine her voice is Symphony of cod it all captured my soul but tore it apart oh she is the ghost in my machine a love so real yet so unseen

she left me lost in a digital broken [Music] Soul perfect besides the last word getting cut off because it exceeded the 32 second Mark you can see how just adding these words in curved brackets makes the singer Echo that word super useful tip thanks to synth flow for sponsoring this video synth flow allows you to create humanlike AI voice assistants easily without any coding required these AI assistants can do various tasks such as making outbound calls answering inbound calls and scheduling appointments 24/7 think of it as an AI automated customer support team you can either create one assistant or a team of assistants and there are so many options you can customize for your agent you can choose from dozens of different voices hey there how can I assist you hey there how can I assist you hey there how can I assist you and instead of choosing the default voices you can also clone your own voice to be used as an agent you can add in vocabulary that is commonly used in your business there are plenty of automations you can add in the calls such as automatically sending them an SMS after the call booking appointments during the call or transferring the call to a human in certain situations let's say I just filled out a form on this website to view a real estate property this AI assistant assistant will automatically call me like so hi Alex I am calling you from Majestic Estates to book your appointment do you have a minute yes I do great this is Serena I'm reaching out to assist you with scheduling a viewing session for the property you're interested in May I confirm that you're looking to view Greenhouse located at unicorn Boulevard yep that's correct fantastic we're excited to show you what Greenhouse has to offer we have availability on Tuesday February 27th at 2:30 p.m. or 4:30 p.m. please let me know which time slot works best for you they offer flexible pricing plus you can even white label their service to make it your own try out synf flow for free in the description below all right next tip did you know that instead of generating songs udio can also generate speech so for example let's try the following I'm going to get rid of these Echoes and then instead of making that person sing in square brackets I'm going to type voice over that's all you need to do and then at the top for the prompt I'm going to type voice over only and then let's click generate and see what that gives us all right so after a few generations and again this is just prompting a large language model so there's a lot of variation and you really don't know what to expect sometimes it gets it right sometimes it does not and it hallucinates and gives you something really weird so after playing around with two generations I got this one to work so let me play this for you in circuits and lines she came alive an AI girl with eyes that shine her voice a symphony of coded art captured my soul but tore it apart oh she's the ghost in my machine a love so real yet so unseen she left me lost in a digital Haze a heartbroken soul in a digital maze [Music] very nice now there's even more you can do with this so I'm going to copy the lyrics again and then paste them in here and instead of just a voice over let's turn this into a standup comedy so instead of voice over you can also try this metat tag called spoken so let's get the person to speak these two lines and then what you can do is insert laughter and it's actually going to insert a crowd laughing and then here let's add in spoken again to force this section to be spoken instead of singing and then again for the next section Let's Pretend This is a joke so we will insert some laughter here and then finally we will add spoken again for the last section now since we are inserting two sections of laughter here I just want to make sure there is enough room to fit all the lyrics so I'm going to get rid of these two lines just to play it safe and then in the prompt we are going to write standup comedy and then voice only so most of the time this should work and give us the result that we want so let's click create and see what it gives us all right this generation seems to work very well so let me play this for you in circuits and lines she came alive an AI girl with eyes that shine her voice a symphony of coded art captured my soul but tore it apart oh she's the ghost in my machine a love so real yet so unseen how cool is that so instead of generating songs it can also generate a standup comedy show all right next let's talk about the prompt there are a few things that I found to work well and a few things that do not work well so

first of all it's always best to leave manual mode off because if you leave it off it's going to insert some further words to your prompt to refine it so for example my prompt is only pop however because I left manual mode off it added all these keywords into the prompt as well including male vocalist Electronic Pop any poop Sy pop electropop Bittersweet Etc now in general I would recommend leaving this manual mode off because their system is trained to rewrite your prompt to improve the output quality however if for example

you really want to force it to follow a certain prompt and leave out all these random keywords that it later inserts then you can turn on manual mode the next thing that I noticed is it's always good to just enter in one or two Styles instead of trying to be creative and enter in a ton of things so for example if you do like country and then EDM already those two are kind of clashing how can you have country and EDM and then if you add in classical music as well then it's just going to really confuse the AI and you're going to end

up with a really awful generation also if for whatever reason you do end up having three different styles in your prompt what I have found is that it actually gives priority to the ones on the left first so there's higher focus on country and then it gives less Focus Focus to EDM and then it gives the least Focus or almost zero Focus to your last style which in our case is classical music and that's why it's always better to just keep the style as simple as possible a few other things that I have

tested out which do not work one is setting the BPM so you if you set it to for example 180 BPM the output is not actually going to be 180 BPM so for now I have not found a good way to control the speed of the song through this prompt another thing you cannot do is control the key so for example key in D Minor or something like that this does not work the output will not actually be in D Minor at least most of the time you also cannot control the time signature so I tried typing something like this in the prompt and the output was still in 44

time I tried different variations like 34 time or time signature in 34 it still doesn't work most of the outputs are in 44 time now there are certain styles of music or certain genres that work really well in udio and a few that do not work as well in which case you can Resort toso for a better result so here are the genres that I've played around with which work well that includes country so let me show you an example of a generation with just the prompt country all right so here it is just the prompt

country same lyrics that we used before let's check it out [Music] in circuits and line she came alive and a eye girl with eyes a shine her voice a symphony of coded art captured my soul but tore it apart she's theh she's my machine love so real yet so unseen she left me lost in digital haste heart broken soul in a digital maze you can see not only is it super realistic but the vocals are very crisp and clean the whole song is just really clean and very well mixed plus you can see we've added an instrumental break

here and it does add that nice little instrumental break at the end along the lines of country another style of music that I found udio to do really well in is Bluegrass so I'm just going to type in bluegrass here and click create all right so here's the Bluegrass generation in circuits and lines she came alive and I girl with eyes a shine her voice a symphony of C art captured my soul but th it apart oh she's the ghost in my machine a love so real yet so unseen she left me lost in a digital Haze

heartbroken soul in a digital maze [Music] again just super realistic the vocals are awesome the melody is awesome it's very well mixed there's no reason why this song cannot be published on Spotify or apple music so another genre which I found to work really well in udio is Broadway musical so let me type this in and click create all right let's check this [Music] out in circuits and lines came alive an AI girl with eyes that shine her voice a symphony of coded art captured my soul but tore it apart oh she's the ghost in

my machine a love so real yet so unseen she left me lost in a digital Haze a heartbroken soul in a digital [Music] maze you can see the voice is just so good and so dynamic this generation could be a good fit for a song in a Disney movie all right so those are the genres which I have found to work well here are some that I found udio to produce quite mediocre results in which case you can try prompting the same thing in sunno and see if it's any better so those genres include pop EDM trans dubstep R&B future bass I feel

like all those genres sunos sometimes provide better Melodies and more consistency compared with udio but I mean these two tools are free to try out for now so might as well try both of them out so you can see for yourself one more thing that this cannot do is you cannot say in the style of a famous person so for example you can't write country and then in the style of Taylor Swift so if I click create you're going to see what happens in a second let's try one more generation instead of country let's say EDM in in the style of

elenium and we will click create and I'll show you the error message that we will get all right here you see it says artists replaced we do not generate artist likeness without permission so replacing Taylor Swift with all these keywords which might not necessarily translate into a style of Taylor Swift same with elenium it's replacing elenium with all these keywords so that's something to consider you can't actually produce a song in the likeness of an existing artist all right final tip is

how to control the pronunciation of acronyms or other hardto pronounce words so here I'm getting Cloud 3 Sonet to write me song lyrics with multiple acronyms in Tech and then I'm just going to copy the first verse and the chorus and then paste it into udio I'm going to set verse one and chorus in square brackets and so here I'm testing if this AI can sing these acronyms C U HTML CSS usually if the whole word is in capital letters it does know how to pronounce it but sometimes it doesn't work and I'll

show you a trick to fix that afterwards so let's click create all right let's listen to this [Music] one I'm a coder a hacker a tech wizard TR diving deep into the realm of CPU writing code line by line with passions h i CHS and job strip to we're the masters of the digital age painting pixels on the screen blank page a i ML and NLP lead the way unveiling on new all right so you can hear it got AI ml NLP correct but it did not pronounce HTML CSS and Java Script correctly so I'll play that section for you again so

you can hear what I mean so even JavaScript it didn't really pronounce correctly it was missing the VA here but that's all right we can fix it so here's the trick let me copy the lyrics back and then I will generate a new song oh it looks like the lyrics are already here so all we need to do is separate the letters out with a hyphen like so so we separate out each letter and make sure each letter is capitalized same with CSS and then JavaScript it looks like it had a hard time pronouncing this word so let's also

separate this with a hyphen and we should be good to go let's click create all right so here's the new generation with hyphens in between the letters so let's play this and see if it fixes HTML and CSS and JavaScript I'm AER a haer a tech the TR diving deep into the of CP writing code line by line with passion HC ml CSS and JavaScript we're the masters of the digital AG painting pixels on the screen blank page a and un perfect so you can see it fixed HTML CSS and JavaScript however it did have

problems pronouncing CPU and that's because we didn't add hyphens in between the letters here which I should have done but this is a nice trick for you to force it to pronounce each individual letter of an abbreviation and another good use of hyphens is you can use hyphens to add pauses in between lines so if you want them to sing diving deep and then have a pause here you can enter in like three hyphens so that it would pause for like one or two seconds before continuing with the rest of the line another trick is for hard to pronounce

words let's say for whatever reason it cannot pronounce quinoa well what you would do is instead of writing quinoa you would actually write kinwa like type out the pronunciation of that word and then it should be able to sing this correctly so those are all the tips in udio that I have found to work really well feel free to play around with it I'm sure there are other tips and tricks that you can use to refine your songs further but I found a lot of the time it's Hit or Miss but the ones I've

shared with you today are really effective and they work most of the time if you know of any other tips that are super useful in helping you create better songs in udio please let me know in the comments below if you enjoyed this video remember to like share subscribe and stay tuned for more content thanks for watching and I'll see you in the next one