Thoughts on Suno V4? If someone wanted to make AI music for the first time this is my recommendation rn. Talks are that a third player has something cooking that might be better...
I'll bet it's a China-based service. Not having to deal with copyright infringement issues goes a long way toward building a robust generative model and China has zero regard for US and international copyright laws.
Dude is brimming! Man its so fun to watch your videos and see the new things coming out! Keep up the great work Matt! We all really appreciate your effort and time on these projects and showing this new stuff! Wow.
I wish the Personas captured just the vocals (or a specific instrument). That's my biggest issue with this model, there's no way to edit an instrument individually. It would be great is it broke out separate channels for each instrument or vocal. It also doesn't do multiple vocalists, choirs, etc very well, you can choose between male or female vocals, but there should be some means of annotation in the lyrics like / / / etc. They need some sort of markup. Also lyrical timestamps would be nice. A means to set tempo. It won't do dissonant chords at all. The simpler prompts work better. And it leans WAY WAY too far toward pop influence, especially with female vocals. This is good for people posting RUclips videos who need background music, but it offers almost nothing to an actual musician. You have to run prompts a million times to get anything usable. Waiting for Suno to come up with a hit is like the proverbial monkey with a typewriter producing Shakespeare. I do see promise in it, but they keep adding features on top of a model that fundamentally needs to be rebuilt from the ground up.
Dude is brimming! Man its so fun to watch your videos and see the new things coming out! Keep up the great work Matt! We all really appreciate your effort and time on these projects and showing this new stuff! Wow.
I‘m not sure if that is an ironic post or do you really have the feeling of authentic emotions… I really love the information I get from Matt, but all his mimics and reactions feel so unauthentic and acted for me. Like he saw that somewhere else on RUclips and thinks he needs to do that also.
VERY impressed. Even when you were saying "ooh, not very good" I was like "What?!" -- I'm just so blown away we're already this advanced in ai music generation. Even the lyrics are INSANELY good in some of these songs, imo. Can you imagine where we'll be two Sunos down the line?
Ya… at some point in the last few months it felt like this systems were getting worse. But it’s picking up again. Improvement comes in waves it’s seems like…
I love these AI music update videos with the new versions it is mind-blowing how good it is getting especially that remaster of the Southern Christmas song just amazing bro 16:29
Vocals are so much better on V4. Removing a lot of that underlying fuzz/distortion on the vocals. Guitar solos still sound like instruments other than guitar. I've personally been working on depressing ai songs some ai generated lyrics and some written by me to do something a little different and love this update.
I agree this new update is the chisel, they have improved on many aspects that needed addressing. I’m impressed with the first output. I did keep it up Suno!
Udio still blows Suno away, IMO. Suno suffers from some kind of high end noise and distortion which really grates on my ears. It might be whatever compression algorithm they're using. Udio does not have these artifacts or distortion in the final mix. I do like their Personas addition. This is a good quality of life improvement.
@@luizhenriquepaes8991 Nothing mind-blowing? It produces better music than 99% of musicians. It's amazing how quickly some people become jaded by AI technology. 🤭
@@luizhenriquepaes8991 Yes it is the autotune-like vocals which will probably make me stick with Udio. Suno does have a lot going for it in other ways though. The song structures and progression are more robust, where Udio goes off on wild tangents that have to keeep being edited out and re-tried.
A big improvement, personally in the past I’ve always much preferred Udio as you can hear the lyrics better, but this seems like something they’ve fixed
As a Suno user, I feel V4 has better sound quality over V3.5, but V3.5 is still more consistent in speech recognition when creating personas. It's improving though. Great video!
25:30 I mean, that makes sense they probably just avoid as much as possible you being able to take someone's existing invoice and use it without their permission
I think we will look back and say we had to do prompts? because in the not to distant future we just talk to the AI like we would talk to professionals in the studio, like soon we will have multiple agents we just tell in natural language exactly what we want. We will easily be able to interact and modify parts of the song, change tempo, ask for new vocals, add instruments, no different than we do in life now, just instead of people, it will be AI agents. And yeah the same for video, just talk to the AI like you would a cameraman, sound man, effects, lighting, etc.. We will talk to teams, crews, those AI teams will also talk to each other to get the project done. Right now the interfaces are not great, talking to a single AI is also going to change for all work, a single dedicated AI that understands lighting or effects and just collaborates with everyone brings about better results and easier work flows. AI will work exactly like a true team and pull off much better results when each agent is focused and trained on a particular area. I can't dream what 2 years will bring outside of the fact it will be massively different than we even imagine.
I'm not sure.. That sounds as if a lot of people will no clue will act as architects without knowing architecture so the resulting house will be a mess. The creation of songs from a simple prompt means not having to micromanage every note or sound.
In high end audio the concept of the 'silence between notes' has always been a big thing. This version gets closer to the ideal. Less 'groove noise' and hash in the vocals. I figure version 7 will be 99% perfect. Kudos to Suno for chasing the dream. Thanks for your measured enthusiasm. ❤
I've spent 30 minutes with it. And yeah, the vocal quality is noticeably better and improved compared to v3.5. *Edit after fiddling with it for a few hours:* You may find it buggy and run into issues, especially if you are using the new "remaster" feature. My suggestion would be to wait a day or two before blasting your credits. Don't do what I did 😄 Understandably, the server demand is very high right now, and I couldn't gauge how much of the issues I was running into are due to the new update or simply traffic related.
That first song was pretty lit. I still remember the video you did on the amazing new "song" generator that basically made static-filled beeps and boops. That must have been months ago. Good times.
The idea that songs you've spent time and effort on in the past can be remixed and remastered easily is a great way to keep people coming back and using Suno. I used to use Udio back in the day, but I haven't seen their long-standing issues worked out yet and it takes so much more effort to get something usable. I'm hoping for more competition soon, but Suno has been great so far.
@@jasonshere I really like Suno because it has lots of features that Udio doesn't have. The only issue I have with it is the audio quality. After a few seconds of listening, I can tell that it was made with Suno.
From my tests, here's what I think: the vocal quality and structure variance are much improved, but also there seems to be less variety in the vocals. Like, every song in a melodic genre (i.e. not rap) with female vocals is a little slower and a little quieter than expected, even when you ask for a genre like metal that's typically louder. I kind of feel like the best approach going forward is to just generate with v3.5 to get something you like, then remaster to v4 for the quality boost. It also still struggles with duets... though that may be because I pushed it too far by testing "male rap, female singing" nu metal duets 😅
This all brings me back to my early days as a sound engineer in the studio, during the monumental transition from analog to digital. That shift was revolutionary: recordings became sharper, more precise, and sometimes even better. But something essential was lost in the process. A flawless recording doesn’t guarantee that the soul of the music is preserved. In the pursuit of technical perfection, emotion, composition, and arrangement were often left behind. Today, with the rise of 4.0 technology, this issue feels even more pronounced. Sure, voices sound pure, and arrangements are impeccably polished, but much of the work lacks depth and heart. These are voices I would never have recorded in the studio back then: they don’t move you, they carry no weight, and they tell no story. The result? Polished perfection without soul-and that might be the greatest tragedy of all. Take Suno AI, for example. I was an early adopter and admirer of their work. But recently, they seem to have lost their way. The results are technically superior but compositionally weaker. Clicks, strange tones, even moments where the AI clearly struggles: is this due to technological limitations or the influence of record labels pulling strings behind the scenes? As a former a&r producer for a label, I know how intense the pressure to deliver can be. But the difference between a track worth releasing and one destined for the shelf is undeniable. Much of what now emerges on 4.0 platforms would never have been released back then. I would take an imperfect recording on an old cassette over a lifeless studio masterpiece any day. A great composition can always be re-recorded, refined, and revived. But a soulless track? It’s dead on arrival. 4.0 developers must take this to heart: artistry must always take precedence over technical refinement. For professional users, this is a crucial truth. Perhaps the average home listener doesn’t notice it immediately, but for those of us who care deeply about the soul of music, this direction feels like a step backward. Companies like Udio may deliver perfectly produced recordings, but they lack the soul that once set Suno apart. Even Suno’s recent work feels like a shadow of what it once was. Let’s hope they find their way back. Because without emotion, without stories, all that remains of music is an empty echo.
There is no such thing as a soul… only the mind. And this machines are simulating the mind in some form. If it lacks “soul” that just means that the simulation isn’t good enough. Soon enough it will be tho. And eventually, this systems will have a “soul”, also. A perfect simulation is no different from the real thing.
The improvement in production quality is very notable, but most of the demos don't feel like they have a human touch, at least to me. Definitely impressive by AI standards, but not at the level where I feel that it competes with many professional musicians. The last song was kind of scary though. Surprisingly close to the general level of quality of indie/rock radio, and only subtly lacking in dynamics and vocal phrasing. Yes, it's generic, but a lot of popular music is, so that lowers the required bar to reach.
We are approaching the existential moment in A.I where we question the uniqueness of Human creativity but to be fair these songs are being built off of human creativity.
And so it will ever be human, that's good, it's like our child. Unless it becomes advanced to the point it can gather it's own new data autonomously...
I'm really impressed with the new model. While the remaster can sometimes be finnicky and mess up the original melody, the quality of the model is a big upgrade and very clearly the top dog in the AI music space.
Suno still has that overexposed, robotic sound to the vocals, so I'll wait for v10, lol. It's definitely not as much as before, but still bad. It definitely bothers me. It's been like that for all AI, though. I can't call it art just yet, but it will get there.
I got very little out of sumo, I wanted it for epic soundtrack for film. It kept creating beats etc. I think udio is better for out of box stuff. Or maybe i was doing it wrong in suno?
AI-generated music is entering an era poised to replace human dominance on the Billboard charts, showcasing immense potential for enhancing its quality.
It would be cool to attach such a plugin to FL to use specific compositions, specific musical instruments, create samples, etc. It would just blow your mind!
I accidentally listened to several songs on the main page and there is no noise and distortion! The quality is almost like that of Udio, even written in version 3
Ok this is awesome, surprisingly so. The real question is, as a guitarist, can I feed it a track and have it generate the rest of the band and then iterate on changes? That’s the holy grail.
Well, what a pleasant surprise. Hello to you sir. It prides me to not only fortuitously encounter you, but also to now receive the opportunity to express my gratitude for the subscription worthy content you provide. Please, be well.
Well I still clearly can say which song was generated with Suno and which one with Udio :). There's still a specific noise in Suno's vocal... while Udio doesn't have that at all. But V4 has nice song construction, pretty --> DAMN
I haven't seen that many people talking about it here in the comments, but I have noticed some major problems with V4. Most noticeably in the beginning of songs, its some weird artifacts or wobbly ringing sound on every single song I make.
While I was able to successfully create Greek songs using the previous version v3.5, I have encountered significant issues with the new version v4. Specifically, v4 appears to lack Greek language recognition. This results in frequent misspellings of Greek lyrics by the singers in generated songs. This problem was infrequent in v3.5, but it is consistently occurring in v4. As a Premier Plan subscriber, I believe that v4 seems unsuitable for creating Greek songs. I will continue using v3.5 and hope that future updates to v4 will address the Greek language issue, preventing such lyric errors in the future.
I been playing around with V4. My experience is it feels very alpha-beta and a lot more miss then hit then v3.5 with lyrics and quality. I am sure it will get better with improvements in the future. I gone through about 2000+ credits using v4 and only about 40% was the quality shown in this video. Lyric prompts have improved. example; typing in CAPS makes the vocals sing loud/yell in that part of the song now
The one thing I wanna add is that artists when singing they it doesn't sound like their voice. Many artists in these TV singing shows when they start speaking some of them sound the same but some it's like someone else. So I think that if we don't hear our voice we will believe it's not us (since most of us do not have a music background)
How can You be emotional for your song that AI wrote? You were giggling and thinking of silly prompts while AI was copying the sweat and tears of musicians that wrote songs like this and took some time, maybe years, to process the emotional turmoil and fit an song around it. Now that its written to paper we must perform it over and over and finally record it, which takes two days to a week. All the while, You have clicked a button
It’s called advancement. Work smarter, not harder. Also, welders and machinists need decades to master their work. Yet no one is singing their praises. And they are far more difficult jobs (physically and mentally) than singing/writing. It is more dangerous and less celebrated, yet far more essential. But, no one gives a fuck about em. Do not not bitch to me about your sweat and blood. Everyone sweats and bleeds for their money. As did the people who make this models. Musicians have it quite easy. This is advancement. Also, no… AI wasn’t copying anyone same as artists aren’t copying each other. This TOOLS use data acquired from people’s work to cross reference certain words and phrases with objects, sounds and images. It identifies the patterns and structures within existing data to generate new content. THAT’S WHAT HUMANS DO EVERYTIME THEY THINK! We cross reference everything relevant we have stored in our brain everytime we think(or things based on that relevant data)… that includes music. We do the same thing as this models and it’s not copying or stealing. So AI doing the same thing as us, is also not copying or stealing.
It’s cool but I still think I prefer Udio’s vocals, I find Suno’s still has a slight robotic sound to them, especially when voices sing in higher registers. Also found that V4 often tries to add “build ups” and “drops” more frequently than before for many genres
As with the previous version of Suno, I’m hearing choppiness in the vocals. I think the current version of Udio may still be better. I’m waiting for the next version of Udio.
Thoughts on Suno V4? If someone wanted to make AI music for the first time this is my recommendation rn. Talks are that a third player has something cooking that might be better...
Third player you say? "Quietly prays it's ElevenLabs"
Song of the year Super Soup
I'll bet it's a China-based service. Not having to deal with copyright infringement issues goes a long way toward building a robust generative model and China has zero regard for US and international copyright laws.
Dude is brimming! Man its so fun to watch your videos and see the new things coming out! Keep up the great work Matt! We all really appreciate your effort and time on these projects and showing this new stuff! Wow.
I wish the Personas captured just the vocals (or a specific instrument).
That's my biggest issue with this model, there's no way to edit an instrument individually. It would be great is it broke out separate channels for each instrument or vocal.
It also doesn't do multiple vocalists, choirs, etc very well, you can choose between male or female vocals, but there should be some means of annotation in the lyrics like / / / etc. They need some sort of markup.
Also lyrical timestamps would be nice. A means to set tempo.
It won't do dissonant chords at all.
The simpler prompts work better. And it leans WAY WAY too far toward pop influence, especially with female vocals.
This is good for people posting RUclips videos who need background music, but it offers almost nothing to an actual musician. You have to run prompts a million times to get anything usable. Waiting for Suno to come up with a hit is like the proverbial monkey with a typewriter producing Shakespeare.
I do see promise in it, but they keep adding features on top of a model that fundamentally needs to be rebuilt from the ground up.
IMPORTANT: For better audio quality download the wav version of the song. The one played on the website is always reduced quality!
Who would have guessed that
Udio does not have this problem. Both the streaming and downloaded versions are very high quality.
@@vomm it's for bandwith saving I guess
@@ManHobbyChannel Oh who would have guessed that. You're smart 🤗
Have you found any other Ai services that further enhance the sound quality after downloading? just curious
Dude is brimming! Man its so fun to watch your videos and see the new things coming out! Keep up the great work Matt! We all really appreciate your effort and time on these projects and showing this new stuff! Wow.
Thanks a ton!
It's incredible how far AI song generation has come since just a year or two ago
Yeah And this is just V4, who knows what would be like if V5 has been announced
@@NoahtheGameplayerSuno 10.0 🎉 wow
@@NoahtheGameplayernext gen udio model will be it create whatever the hell you can
Time moves differently in the AI world, Imagine having 100 years to perfect 1 thing.
I don't know which is more entertaining... the songs or your reactions. Thank you for this--I needed it today!
Glad you enjoyed!
this has been one of the greatest tools ive ever had for content creation
Suno is always easily recognizable because of that vocal quality that's like having a chorus effect in 100%.
I so love the fun you have while listening to the generations! So real!
Same!
I‘m not sure if that is an ironic post or do you really have the feeling of authentic emotions…
I really love the information I get from Matt, but all his mimics and reactions feel so unauthentic and acted for me. Like he saw that somewhere else on RUclips and thinks he needs to do that also.
VERY impressed. Even when you were saying "ooh, not very good" I was like "What?!" -- I'm just so blown away we're already this advanced in ai music generation. Even the lyrics are INSANELY good in some of these songs, imo. Can you imagine where we'll be two Sunos down the line?
Ya… at some point in the last few months it felt like this systems were getting worse. But it’s picking up again.
Improvement comes in waves it’s seems like…
I love these AI music update videos with the new versions it is mind-blowing how good it is getting especially that remaster of the Southern Christmas song just amazing bro 16:29
Im hyped for suno-4-turbo and suno-4o. And suno-o1 will be a game changer for sure
Naw it will never be released. Still waiting on the text based music version of suno they have been teasing for a year.
Nah, we'll have to spend a few months on suno-o1 Preview first.
It's perfect for me, someone with no musical talent.
Same! 😂
I've been listening to Udio music more than spotify these days. Amazing tunes.
Same here :)
Really? I’ve gotta check that out then!
because this industry is monopolized by players who serve us the same recipe over and over again, provided that AI fucks their sectarian system well
That's sad
Totally. And it’s better than Suno, at least Suno pre v4
Excellent Updates!!! I still use the 3.5 Free Plan, but may upgrade some day. Thanks, Matt, and I love the songs you've created with Suno. Cheers
Vocals are so much better on V4. Removing a lot of that underlying fuzz/distortion on the vocals. Guitar solos still sound like instruments other than guitar. I've personally been working on depressing ai songs some ai generated lyrics and some written by me to do something a little different and love this update.
I love this tool even as someone with musical talent. It is cool to mess around with.
I got hit with V4 just as I entered Suno, I'm speechless. Amazing update!
holy shet. This has stepped up a notch
I agree this new update is the chisel, they have improved on many aspects that needed addressing. I’m impressed with the first output. I did keep it up Suno!
Yo, that was amazing. Absolutely loved it compared to what I’ve heard in the past. We are definitely on the right direction.
Udio still blows Suno away, IMO. Suno suffers from some kind of high end noise and distortion which really grates on my ears. It might be whatever compression algorithm they're using. Udio does not have these artifacts or distortion in the final mix.
I do like their Personas addition. This is a good quality of life improvement.
Exactly!
You can still hear the Suno characteristic sound, it's like it's recorded with a bad mic or a poor equalization.
Plus heavy autotune
It is much better than V3, but nothing mind blowing
but that's the character of it and it works
Yup, aside from those... the plucky and punchy sounds 💀
@@luizhenriquepaes8991 Nothing mind-blowing? It produces better music than 99% of musicians. It's amazing how quickly some people become jaded by AI technology. 🤭
@@luizhenriquepaes8991 Yes it is the autotune-like vocals which will probably make me stick with Udio.
Suno does have a lot going for it in other ways though. The song structures and progression are more robust, where Udio goes off on wild tangents that have to keeep being edited out and re-tried.
Voice cloning might be difficult for them considering the legal issues with singers, but I think I heard of being able to swap vicals with Replay.
From this point on all I would really like are UI improvements and better editing features.
A big improvement, personally in the past I’ve always much preferred Udio as you can hear the lyrics better, but this seems like something they’ve fixed
There are lot of problems i seen on reddit. Done over 200 different song with 3,5 so going to be fun remaster some of them
the community ones are amazing you have built a great community :)
As a Suno user, I feel V4 has better sound quality over V3.5, but V3.5 is still more consistent in speech recognition when creating personas. It's improving though. Great video!
26:04 This stil sounds like the typical bad quality Suno sound :(
Yeah, I don’t understand why they can’t fix this issue.
This site gave me inspiration to make music again. I cried for happiness when I tried it for the first time.
25:30 I mean, that makes sense they probably just avoid as much as possible you being able to take someone's existing invoice and use it without their permission
I think we will look back and say we had to do prompts? because in the not to distant future we just talk to the AI like we would talk to professionals in the studio, like soon we will have multiple agents we just tell in natural language exactly what we want. We will easily be able to interact and modify parts of the song, change tempo, ask for new vocals, add instruments, no different than we do in life now, just instead of people, it will be AI agents.
And yeah the same for video, just talk to the AI like you would a cameraman, sound man, effects, lighting, etc.. We will talk to teams, crews, those AI teams will also talk to each other to get the project done. Right now the interfaces are not great, talking to a single AI is also going to change for all work, a single dedicated AI that understands lighting or effects and just collaborates with everyone brings about better results and easier work flows. AI will work exactly like a true team and pull off much better results when each agent is focused and trained on a particular area. I can't dream what 2 years will bring outside of the fact it will be massively different than we even imagine.
I'm not sure.. That sounds as if a lot of people will no clue will act as architects without knowing architecture so the resulting house will be a mess. The creation of songs from a simple prompt means not having to micromanage every note or sound.
This is like Gpt4 moment, I am so happy
In high end audio the concept of the 'silence between notes' has always been a big thing. This version gets closer to the ideal. Less 'groove noise' and hash in the vocals. I figure version 7 will be 99% perfect. Kudos to Suno for chasing the dream. Thanks for your measured enthusiasm. ❤
I've spent 30 minutes with it. And yeah, the vocal quality is noticeably better and improved compared to v3.5.
*Edit after fiddling with it for a few hours:* You may find it buggy and run into issues, especially if you are using the new "remaster" feature. My suggestion would be to wait a day or two before blasting your credits. Don't do what I did 😄 Understandably, the server demand is very high right now, and I couldn't gauge how much of the issues I was running into are due to the new update or simply traffic related.
This is great can’t wait to check it out
I have been playing with Suno V4 since 2 hours after it came out today and I am truly impressed!
i love how you love this as much as i do!!!
It does best using an LLM for the lyrics and then use custom and pick a style using the LLM lyrics. I got some mind blowing results
That first song was pretty lit. I still remember the video you did on the amazing new "song" generator that basically made static-filled beeps and boops. That must have been months ago. Good times.
The idea that songs you've spent time and effort on in the past can be remixed and remastered easily is a great way to keep people coming back and using Suno. I used to use Udio back in the day, but I haven't seen their long-standing issues worked out yet and it takes so much more effort to get something usable. I'm hoping for more competition soon, but Suno has been great so far.
It is all about the right prompting. I am very happy with Udio
@@ahmedkagabo I've made great tracks with Udio; but I can make more faster with Suno. Glad you're happy with what you use.
@@jasonshere I really like Suno because it has lots of features that Udio doesn't have. The only issue I have with it is the audio quality. After a few seconds of listening, I can tell that it was made with Suno.
Audio quality is still second-rate. Kind of sounds like it's playing out of a toy speaker. The music quality is improved, though.
Dude! I can’t wait to go back and remix my old V2 tracks!
Much better quality and i thought the last one was super good! I can't wait to get my hands on it! When's it come out?
I don't care much for the vocal stuff, the stuff at the 30min mark was very impressive and inspiring for me as a music producer.
Wow. I may have to get a subscription to that!! This is awesome!
The singer at around 4:15 sounds JUST like Sam Tsui, I wonder where Suno gets the voices from?
Matt stoping the music at 5:00 had me shouting "NOOO" like Rachel on Days of ourslives set
Frog song was my favorite. It was hilarious.
Will those who don't have a subscription be able to use this version?
No you'll get a free 1
That’s sick man!!! Hey, what is the genre combo prompt and do you have any secret prompts like [Verse 1: Chipmunk] or something ? For DNA, sorry
I remastered all my best songs and I am just blown away at the clarity. I love it! Best upgrade so far.
i have been waited for this
From my tests, here's what I think: the vocal quality and structure variance are much improved, but also there seems to be less variety in the vocals. Like, every song in a melodic genre (i.e. not rap) with female vocals is a little slower and a little quieter than expected, even when you ask for a genre like metal that's typically louder. I kind of feel like the best approach going forward is to just generate with v3.5 to get something you like, then remaster to v4 for the quality boost.
It also still struggles with duets... though that may be because I pushed it too far by testing "male rap, female singing" nu metal duets 😅
All you need is a shredding solo mode!!!! Guitar-Drums- Bass!!!!!!! SUMO REQUEST !!!!!!!!!
for suno to get your voice copied better, upload a 2min wav clip. works like a charm. made me sing like an angel.
also it helps to talk not try to sing, that is when the AI just totally messes up haha.
That "Pass the Butter" song slams.
This all brings me back to my early days as a sound engineer in the studio, during the monumental transition from analog to digital. That shift was revolutionary: recordings became sharper, more precise, and sometimes even better. But something essential was lost in the process. A flawless recording doesn’t guarantee that the soul of the music is preserved. In the pursuit of technical perfection, emotion, composition, and arrangement were often left behind.
Today, with the rise of 4.0 technology, this issue feels even more pronounced. Sure, voices sound pure, and arrangements are impeccably polished, but much of the work lacks depth and heart. These are voices I would never have recorded in the studio back then: they don’t move you, they carry no weight, and they tell no story. The result? Polished perfection without soul-and that might be the greatest tragedy of all.
Take Suno AI, for example. I was an early adopter and admirer of their work. But recently, they seem to have lost their way. The results are technically superior but compositionally weaker. Clicks, strange tones, even moments where the AI clearly struggles: is this due to technological limitations or the influence of record labels pulling strings behind the scenes? As a former a&r producer for a label, I know how intense the pressure to deliver can be. But the difference between a track worth releasing and one destined for the shelf is undeniable. Much of what now emerges on 4.0 platforms would never have been released back then. I would take an imperfect recording on an old cassette over a lifeless studio masterpiece any day.
A great composition can always be re-recorded, refined, and revived. But a soulless track? It’s dead on arrival.
4.0 developers must take this to heart: artistry must always take precedence over technical refinement. For professional users, this is a crucial truth. Perhaps the average home listener doesn’t notice it immediately, but for those of us who care deeply about the soul of music, this direction feels like a step backward. Companies like Udio may deliver perfectly produced recordings, but they lack the soul that once set Suno apart. Even Suno’s recent work feels like a shadow of what it once was.
Let’s hope they find their way back. Because without emotion, without stories, all that remains of music is an empty echo.
There is no such thing as a soul… only the mind. And this machines are simulating the mind in some form. If it lacks “soul” that just means that the simulation isn’t good enough. Soon enough it will be tho. And eventually, this systems will have a “soul”, also.
A perfect simulation is no different from the real thing.
The improvement in production quality is very notable, but most of the demos don't feel like they have a human touch, at least to me. Definitely impressive by AI standards, but not at the level where I feel that it competes with many professional musicians.
The last song was kind of scary though. Surprisingly close to the general level of quality of indie/rock radio, and only subtly lacking in dynamics and vocal phrasing. Yes, it's generic, but a lot of popular music is, so that lowers the required bar to reach.
We are approaching the existential moment in A.I where we question the uniqueness of Human creativity but to be fair these songs are being built off of human creativity.
And so it will ever be human, that's good, it's like our child. Unless it becomes advanced to the point it can gather it's own new data autonomously...
Finally! with V4 i can finally reimagine my song i made back with suno was still only in discord
I'm really impressed with the new model. While the remaster can sometimes be finnicky and mess up the original melody, the quality of the model is a big upgrade and very clearly the top dog in the AI music space.
i honestly think you should upload the songs you've generated in the channel, i really like the super soup song. i want to hear the full song.
Suno still has that overexposed, robotic sound to the vocals, so I'll wait for v10, lol. It's definitely not as much as before, but still bad. It definitely bothers me. It's been like that for all AI, though. I can't call it art just yet, but it will get there.
The improvement was great, still, SUNO is light years away from UDIO, specially in vocals
I got very little out of sumo, I wanted it for epic soundtrack for film.
It kept creating beats etc.
I think udio is better for out of box stuff.
Or maybe i was doing it wrong in suno?
AI-generated music is entering an era poised to replace human dominance on the Billboard charts, showcasing immense potential for enhancing its quality.
It would be cool to attach such a plugin to FL to use specific compositions, specific musical instruments, create samples, etc. It would just blow your mind!
suno 대박인거 같아요.ㅜㅜ 하지만 이러다가 어느날 갑자기 Udio가 엄청난 기능으로 짠하고 나타날지도 모르겠네요.:)
I accidentally listened to several songs on the main page and there is no noise and distortion! The quality is almost like that of Udio, even written in version 3
Can you give us a download link for "I Pass Butter?" Amazing song BTW!! :D Great vid!
Ok this is awesome, surprisingly so. The real question is, as a guitarist, can I feed it a track and have it generate the rest of the band and then iterate on changes? That’s the holy grail.
Could either of these make hardstyle/uptempo songs?
Well, what a pleasant surprise. Hello to you sir. It prides me to not only fortuitously encounter you, but also to now receive the opportunity to express my gratitude for the subscription worthy content you provide. Please, be well.
Pass the butter is powerful.
Well I still clearly can say which song was generated with Suno and which one with Udio :). There's still a specific noise in Suno's vocal... while Udio doesn't have that at all. But V4 has nice song construction, pretty --> DAMN
The remastering is no joke!!
will you do a livestream with Suno v4 songs? 😊
Hi, i was curious about the quality of the STEMS you can download now. can you make a video about that?
Been waiting for this
I think you should've given us a Test of 2 Songs of both the models and it be like a minigame to see which one is the Models for V3.5 or V4
I haven't seen that many people talking about it here in the comments, but I have noticed some major problems with V4. Most noticeably in the beginning of songs, its some weird artifacts or wobbly ringing sound on every single song I make.
Yea the same here....really annoying actually......
While I was able to successfully create Greek songs using the previous version v3.5, I have encountered significant issues with the new version v4. Specifically, v4 appears to lack Greek language recognition. This results in frequent misspellings of Greek lyrics by the singers in generated songs.
This problem was infrequent in v3.5, but it is consistently occurring in v4. As a Premier Plan subscriber, I believe that v4 seems unsuitable for creating Greek songs. I will continue using v3.5 and hope that future updates to v4 will address the Greek language issue, preventing such lyric errors in the future.
It let me remaster one song to v4 even on the free plan. It's awesome!
I been playing around with V4. My experience is it feels very alpha-beta and a lot more miss then hit then v3.5 with lyrics and quality. I am sure it will get better with improvements in the future. I gone through about 2000+ credits using v4 and only about 40% was the quality shown in this video. Lyric prompts have improved. example; typing in CAPS makes the vocals sing loud/yell in that part of the song now
The one thing I wanna add is that artists when singing they it doesn't sound like their voice. Many artists in these TV singing shows when they start speaking some of them sound the same but some it's like someone else.
So I think that if we don't hear our voice we will believe it's not us (since most of us do not have a music background)
How can You be emotional for your song that AI wrote? You were giggling and thinking of silly prompts while AI was copying the sweat and tears of musicians that wrote songs like this and took some time, maybe years, to process the emotional turmoil and fit an song around it. Now that its written to paper we must perform it over and over and finally record it, which takes two days to a week. All the while, You have clicked a button
It’s called advancement. Work smarter, not harder.
Also, welders and machinists need decades to master their work. Yet no one is singing their praises. And they are far more difficult jobs (physically and mentally) than singing/writing. It is more dangerous and less celebrated, yet far more essential.
But, no one gives a fuck about em. Do not not bitch to me about your sweat and blood. Everyone sweats and bleeds for their money. As did the people who make this models. Musicians have it quite easy.
This is advancement. Also, no… AI wasn’t copying anyone same as artists aren’t copying each other.
This TOOLS use data acquired from people’s work to cross reference certain words and phrases with objects, sounds and images. It identifies the patterns and structures within existing data to generate new content.
THAT’S WHAT HUMANS DO EVERYTIME THEY THINK! We cross reference everything relevant we have stored in our brain everytime we think(or things based on that relevant data)… that includes music. We do the same thing as this models and it’s not copying or stealing. So AI doing the same thing as us, is also not copying or stealing.
The friend that’s too woke:
It’s cool but I still think I prefer Udio’s vocals, I find Suno’s still has a slight robotic sound to them, especially when voices sing in higher registers.
Also found that V4 often tries to add “build ups” and “drops” more frequently than before for many genres
$10 a month is reasonable for 500 song generations.
As with the previous version of Suno, I’m hearing choppiness in the vocals. I think the current version of Udio may still be better. I’m waiting for the next version of Udio.
wtf, thats a huge improvement, time to use dem leftover credits
Matt ''Pretty Darn Good'' VidPro.
That last song was insane
Massive Frog King sounds very LonelyIsland band style vibes.
Great update
What's my purpose is my new jam🎶😌😂😂
AI music is awesome. Now I'll be able to listen to catchy tunes that actually have positive lyrics.
Can't really say if Suno sounds bad or is it just that your taste in music is super bad.
Love the V4 version the songs sound more better.
thats crazy now i can MAKE my own roadtrip music haha
That image of the Gut Bugs song where someones holding a gun to the gummi bears head
i know you are having fun but it would be nice to see what serious stuff services do