Great video. Let me add some insights from my experience with suno. Structure Tags Tags in the lyrics section like [intro] or [verse] work as well in parentheses if used consistently: (intro), (verse), (interlude), (pre-chorus), (chorus), (hook), (drop), ... [instrumental part] or [buildup] or [drop] seem to work pretty reliably too. Solos [solo] works astonishingly well, but tends to give me just a short solo part like one bar or two. Repeating the tag gives me twice the length. And something like: [reserved solo] [solo intensifying] [intense solo] [solo virtuoso] usually has some really nice longer solo parts. [long solo] works, too, but I found [long intense solo] to not work that reliably. When putting solos in a row it might just change the solo instrument mid-way, though. [long solo part starting reserved and building up to intense virtuoso climax] doesn't work that reliably, but if it works, the results can surprise. Adlibs / backing vocals I found Parentheses around text at the end of a line or stand alone most of the time gets sung as adlibs or backing vocals with a voice / vioces differing from the main singer. You can construct a duet with that but there are timing and other issues when you try. You can get a duet with standard lyrics. Sound Effects Sound FX don't seem to have a consistent syntax to invoke in suno. Some specific FX just need to be named in Asterisks like *Gunshot*. Onomatopoeic or dadaistic expressions mostly work with or without Asterisks. **Peng!** might give you a gunshot, too. Ha-ha-ha will most probably give you laughter. Boom! usually get sung but **Boom!** might give you an explosion.
Very helpful thanks. I'm 23:05 from a music background and found this very frustrating. Especially with the genre and voice style. Sometimes it gives me a completely different genre. Also you are right if you add section where you want changes in music it sometimes sings it or most of the time ignores it. V4 is very frustrating with the unwanted sounds. I don't know who tested this but it was nowhere near ready for release.
It really does have a ways to go before it’s consistent, accurate or able to follow any basic music theory prompts. Sometimes it feels more like spinning a wheel of options and getting surprised by what ever pops up. Sometimes it’s pretty cool, but not always what you were looking for.
Thank you, Matt. I've been using Suno AI to create my own songs for a while now, and you're right. This AI can be quite random in many aspects. The topics you mentioned have really been helpful to me. 🤖
HI Matt, Great video! I picked up some valuable tips from it. I thought I’d share a few techniques I’ve found useful that you might appreciate. Using metatags within a song can allow you to include multiple singers or even a speaking voice in a single track. From my experience, placing the tags at the beginning of a verse (with verses capped at 6-8 lines) works best for clarity. Additionally, you can incorporate a variety of sound effects simply by specifying the type of sound in brackets. Below are some examples that have worked well for me, though there are plenty of other sound effects to explore: Lyrics box: [spoken word] [male vocals] [female vocals] [siren] [screams] [laugh] [clapping]
Thanks for your thoughts and experience with this! I have found less luck than you at trying those techniques within the lyric section, but Suno does seem to be adjusting every day so it might be that it’s starting to do a better job of following those prompts. When I’ve tried those in the past, it’s just ignored them, but I’m glad they’ve been working for you! Hopefully it’ll become more consistent so that we know what to expect and can better make what we want! Thanks for letting me know and keep sending any prompts you find are working!
I Loved the video! Instant sub!. Also!! Look into using ALL CAPS / Title Caps / no caps with [metatags] or in the Style Of Music section. This way you can emphasize the importance / weight over the others. As found in the sunoaiwiki: "In Suno AI, the letter case (upper, lower, or title case) used in prompts can significantly impact the quality and style of the generated song. Different parts of the prompt react differently to letter case styles, affecting the output’s coherence and alignment with user expectations. Solution: To get the best results from Suno AI, use different letter cases for different parts of your prompt. Here’s a recommended approach: ALL CAPS for genres: This gives genres the highest priority. Title Case for descriptors: This gives descriptors a middle priority. lower case for instruments: This ensures instruments are included prominently in the song." Hope it helps!
Another fun one I’ve got a preacher before offer recording my friend reading off from a self-help book and it was literally just like preaching in different voices. Every time I made a song complete a cappella no music something fun to play around with
I have read your comment 5 times and I still dont undersgand what your saying... Is thia cypher , is there 2 conversations going on at the same time? or is it a translation issue? I must know..What is the secret behind your comment,
@@realmattdean If you want more male vocals then disliking your generated songs with female vocals will help tailor more songs being generated with male vocals. Also the entire prompt itself plays a part in which vocal is generated, meaning, just writing "male vocals" won't necessarily do it alone, especially if the prompt for example is "male vocals, pop, emotional, synth-pop", Suno will will 9 times out of 10 give you a female vocal based off of the "pop'' part regardless. Sometimes adding the exact genre you want last in the prompt can fix this issue. Most importantly though is going through and tailoring your profile with likes and dislikes on your songs. After a while you won't even need to worry about the prompt structure anymore and you'll still get the desired vocals 90% of the time. This works with specific voices as well. I've tailored my profile so much that almost all of my male vocals are pretty much the same voice now. Suno support team has also confirmed everything I've just said via emails I have with them.
This video is particularly interesting for the commentary on what Suno AI does not know how to do. At this time (Jan 2025) AI is excellent at some things but does not yet "get" certain concepts that humans manage routinely. I think that is the issue Matt is referencing at the beginning of this video. I truly love what AI is learning to provide for us, and I know it is in its "infancy". And, I appreciate the angst musicians, artists and other feel when they consider their career in the face of the looming AI juggernaut. However, Matt is giving valuable insight. The key to remaining relevant in this era of AI is to know the "ways of thinking" that are core to human behaviors. That means knowing music theory - which is the key to music complexity, novelty and interest - which is what will set human music apart from AI music (at least, that's how it seems to me). Thanks Matt.
Thanks John! So glad you enjoyed it and I'm glad I can be helpful. We're all just learning what these crazy things will do and it's all changing so fast. I think we'll need to rely on each other to find ways to use these to our advantage and not get left behind. Looking forward to hearing what you learn as we lean into the grand future that is being created around us!
would love suno to add when using the cover is that we can choose a persona and also that you can change lyrics, so you can do a cover in another language
Your Mister Roboto seems to be one way to get around Suno's ubiquitous pentameter (?) vocal DNA (architecture, family, sing-song-ie style). Thanx for the tips & tricks, esp. what doesn't work yet. I hope that Suno will update to allow turning either an upload or a derivative of one to become a persona.
Thank you for your thoughts! It’s all growing soooo fast. It feels like by the time I get one video out there are changes but then it still won’t do male or female voices consistently…. 🤷 I’m glad this is helpful for you. Let me know what you find works for you!
If you want more consist vocalist for a song try: [VERSE 1: MALE VOCALIST], [PRE-CHORUS 1: MALE VOCALIST] , Etc. with Exclude Style: Female Vocalist. I've found this to work almost every time.. However, Suno is not always consist for the order of Male and Female vocalists for Duets. You will get a male and female vocalist but Suno may reverse the order of the vocalist or use which ever vocalist fits the lyric best. I've messed with this a lot... HTH
Thank you! I agree about the inconsistency with Suno. It’s consistently inconsistent. It does seem to be a little better this week with male and female prompts than it was a few weeks ago so maybe it’s getting better. 🤷
I've been experimenting with the use of "," to separate elements inside [metatags]. Also, I've seen some kind of ":" use, EG [Intro: strings, Bulid-Up, In Crescendo]
@@SteveM45 that’s amazing. I’ve been testing the upload feature before I make a video on it. But it seems really powerful. I’ve uploaded several instrumentals and it does and interesting job of adding different instruments or changing a guitar song into horns or piano…
Thank you! I’ve looked into it a little, but I’m actually trying to get a better handle on the ai video services. I’ve heard from some people that udio is really good in some areas. Will see what I can do!
Sorry I missed that. I added some lyric prompts in the comments. Sometimes those get missed when I'm putting these up on youtube. If there is one specific one I missed, let me know and I'll add it. Thanks for watching!
I was using Suno and Udio to generate variations on my original music that I uploaded to their sites. This worked well until recently. Now, the results are completely unusable with very distorted random noises and many other bizarre issues. I wrote to them about this, but they just fobbed me off with some lame excuse. It used to work fine, but now it doesn't work at all. Maybe they have given up on that function which would be a shame. I don't want to make tracks that are just fully AI. I just want a virtual songwriting partner to provide possible variations on my own music.
Great stuff. If I wanted to modulate up a chorus, how would I write the meta tags? [chorus] [modulate up] / [chorus: modulate up] / or just [modulate up] and Suno will recognize it as the chorus?
Matt: great job, I write to you in English but I'm Latin, I make all my songs in Spanish, I think there's already a theme there, the whole page sees it in Spanish and with the original language (English) it's more specific and it does strange things to me. sometimes . Regarding putting a series of chords and having the skynet obey it, I have obtained results like this: [G] + [D] +[F] . in most cases. Have any singers stuttered? I have several, and it is almost impossible to fix it later, the replace works when it wants ahaha. Problems with person: I made a profile and used it in a new song, at one point two verses from the original song appeared in the new song hahaha. another with the extensions: I have an example of a one-minute part with a female singer and in the pair new extensions a male singer appears..... or else change the singers in each extension. Sorry for being the messenger of crazy failures haha Thank you Miguel
@@enifx wow - yes… Suno can be very very inconsistent. Sometimes when I do a persona, it does it as a male, then changes to female. It’s very strange that Suno brought back in the lyrics from an old song when you imported the persona!! Wow. 🤦♂️ let me know what else you find. I think I need to do a funny video next on all the hilarious errors that Suno does!!
It does a voice and instrumental stem, but it’s pretty bad. It strips the song after creating it unlike a normal stem in protools and it’s full of a lot of bugs. I would love a version that could separate the instruments.
That’s a tricky one. I can get Suno to do a few harmonies and a few duets, but it’s pretty random. Call and respond works most of the time… ie: [verse] I love you (you’re the stars in the sky) I love you (you make me fly)
Although time signature requests don’t work, if I randomly (e.g.) get a song in 6/8, I Extend from almost the very beginning and try different styles. Sometimes the results are good and gets me a 6/8 song.
Specifying [odd time signature] has worked for me. One in particular, I got a 7/8 verse and asked for [half time] on the chorus and got a half time 4/4. I also included "progressive" in my song description. It ended up fitting my lyrics so well I downloaded it then asked for [epic guitar solo] extending after the chorus lol. Took 4 tries but got something I was looking for and downloaded that. Took an intro I had created and crossfaded that into song I first downloaded then spliced the ending solo and turned out into a 7 minute Suno masterpiece.
for sound effects u need to be more simple like doing [gunshot] or [scream] like very simple things so try like [bang] or [spit] or something and for speaking voice I do [talk] or [talking] and for female or male voice tend to do [Mrs.Shauna] [Mr.Alex] wich seems to work more accurate mostly, and if they duett u just say [duett mrs.shauna and mr alex] but make sure to add duett to the style promt, and for time just spellit out
I recorded my voice and uploaded it to Suno. Thing is, I don't hear my cloned voice on a generated track, and I have repeated this many times, using my credits! Can you briefly explain what I'm doing wrong? Thank you!
I am working on a new video that will come out in a couple weeks regarding the upload and cover functions. I’ve had some really interesting results from instrumentals, but not from voices or from sample of horns. I’m not sure what’s going on with it yet, but I’ll report back as I discover what does and does not seem to work. I have found that the prompts used are VERY hit and miss with this feature - so I’m not surprised that it’s having trouble with voices.
@elnidogaming I’ve had pretty good success using my own voice, and while it’s not perfect, it sounds more like me compared to the Suno-generated voices. Here’s the process I follow: 1. Upload a sample of you briefly singing something with no background noise, just your voice. 2. Use that audio file to extend a new song. 3. Set the 'extend from' timing to where you want the song to start from your voice sample. If you don’t want the sample itself in your song, you can extend from a later point (e.g., 00:08). To create a gap before the song starts, add [break] in the lyrics box for the first few lines. For example: Lyrics box: [break] [break] [break] Your voice won’t be imitated perfectly, but it’ll be close. You may need to go through several iterations to get the generated voice sounding more like yours. Hope this helps!
[Spoken] has about a 70% success rate for me personally oh, and it could be confirmation bias but for say rock, putting 90s rock or 2007 etc seemed to help me skew towards what I was looking for Also for voice expression I get a lot of utility out of prefixing voice e.g sultry female voice or soft southern accent etc. it seems to drastically shift the voice chosen.
That’s great to know! Thank you. I’m going to compile a list of what prompts others are getting to work so we can all contribute. Will try what you’re doing!
Hi Matt, I added you in Suno. I have also a very Long testing time in suno with different Commands and formating . You can here it in my songs. Very interessting are "Lied ist aus".
@@SteveM45 great! Would love to see what’s working for you. I’m Hoping the more we point out what does and doesn’t work, the more Suno will make adjustments.
Your not 100% accurate on the voices being matched with genre, i have made many songs using other voices, the best way to control it is to use the lyrics section phonetically, most my lyrics are Meta tags and phonetics, it looks almost like code, this way I have had success controling everything from dj scratching to james brown-like voice to a drunk clown....
I’m quite sure I have a LOT to learn about getting this to work properly. That’s amazing that you’ve found some solutions. I’d love to see an example of what you’re talking about and I’m sure others would as well!
@@SteveM45 interesting. That seems to fall into the idea that Suno is just playing by ear rather than responding to specific directions. But it means that we can’t seem to make something like an odd rock beat in 7/8??
Just bare in mind that Suno owns any music you make using their app... if you use it in a RUclips video you will eventually get a copyright strike.... your "new band member" owns the band's work.
Not according to their legal documents. If you are using the paid service, you own the copyright. It’s only if you are using the free version that you do not. However, the copyright issue with ai music is a long way from being settled.
Of course you will get the strike, because suno protects your work to be stolen and uploaded by anybody as their creation. The problem is there seems to be no white list solution yet to allow you to upload your own works to your own youtube account. And there is no other way to prove your rights than your suno account itself. I think we need a digital watermark solution to sign digital works in the files itself to be able to prove ownership.
@@SteveM45 interesting! I tried :30, :60, 1:30…. Nothing seemed to work. Most songs seem to naturally end between 2:30 and 3:30 so I didn’t try those. Thanks for letting us know what worked for you!
Great video. Let me add some insights from my experience with suno.
Structure Tags
Tags in the lyrics section like [intro] or [verse] work as well in parentheses if used consistently: (intro), (verse), (interlude), (pre-chorus), (chorus), (hook), (drop), ...
[instrumental part] or [buildup] or [drop] seem to work pretty reliably too.
Solos
[solo] works astonishingly well, but tends to give me just a short solo part like one bar or two. Repeating the tag gives me twice the length. And something like: [reserved solo] [solo intensifying] [intense solo] [solo virtuoso] usually has some really nice longer solo parts. [long solo] works, too, but I found [long intense solo] to not work that reliably. When putting solos in a row it might just change the solo instrument mid-way, though.
[long solo part starting reserved and building up to intense virtuoso climax] doesn't work that reliably, but if it works, the results can surprise.
Adlibs / backing vocals
I found Parentheses around text at the end of a line or stand alone most of the time gets sung as adlibs or backing vocals with a voice / vioces differing from the main singer. You can construct a duet with that but there are timing and other issues when you try. You can get a duet with standard lyrics.
Sound Effects
Sound FX don't seem to have a consistent syntax to invoke in suno. Some specific FX just need to be named in Asterisks like *Gunshot*.
Onomatopoeic or dadaistic expressions mostly work with or without Asterisks. **Peng!** might give you a gunshot, too. Ha-ha-ha will most probably give you laughter. Boom! usually get sung but **Boom!** might give you an explosion.
Very helpful thanks. I'm 23:05 from a music background and found this very frustrating. Especially with the genre and voice style. Sometimes it gives me a completely different genre. Also you are right if you add section where you want changes in music it sometimes sings it or most of the time ignores it. V4 is very frustrating with the unwanted sounds. I don't know who tested this but it was nowhere near ready for release.
It really does have a ways to go before it’s consistent, accurate or able to follow any basic music theory prompts. Sometimes it feels more like spinning a wheel of options and getting surprised by what ever pops up. Sometimes it’s pretty cool, but not always what you were looking for.
Thank you, Matt. I've been using Suno AI to create my own songs for a while now, and you're right. This AI can be quite random in many aspects. The topics you mentioned have really been helpful to me. 🤖
I’m so glad it’s been helpful! It’s a really fun tool and I’m often surprised buy what it comes up with.
I'm just now learning to work with Suno. thank you so much for this video.
So glad to help!!
@@realmattdean I appreciate you. thnx again.
HI Matt,
Great video! I picked up some valuable tips from it. I thought I’d share a few techniques I’ve found useful that you might appreciate.
Using metatags within a song can allow you to include multiple singers or even a speaking voice in a single track. From my experience, placing the tags at the beginning of a verse (with verses capped at 6-8 lines) works best for clarity. Additionally, you can incorporate a variety of sound effects simply by specifying the type of sound in brackets. Below are some examples that have worked well for me, though there are plenty of other sound effects to explore:
Lyrics box:
[spoken word]
[male vocals]
[female vocals]
[siren]
[screams]
[laugh]
[clapping]
Thanks for your thoughts and experience with this! I have found less luck than you at trying those techniques within the lyric section, but Suno does seem to be adjusting every day so it might be that it’s starting to do a better job of following those prompts. When I’ve tried those in the past, it’s just ignored them, but I’m glad they’ve been working for you! Hopefully it’ll become more consistent so that we know what to expect and can better make what we want! Thanks for letting me know and keep sending any prompts you find are working!
I Loved the video! Instant sub!. Also!! Look into using ALL CAPS / Title Caps / no caps with [metatags] or in the Style Of Music section. This way you can emphasize the importance / weight over the others. As found in the sunoaiwiki:
"In Suno AI, the letter case (upper, lower, or title case) used in prompts can significantly impact the quality and style of the generated song. Different parts of the prompt react differently to letter case styles, affecting the output’s coherence and alignment with user expectations.
Solution: To get the best results from Suno AI, use different letter cases for different parts of your prompt. Here’s a recommended approach:
ALL CAPS for genres: This gives genres the highest priority.
Title Case for descriptors: This gives descriptors a middle priority.
lower case for instruments: This ensures instruments are included prominently in the song."
Hope it helps!
Interesting… I hadn’t thought of upper and lower case having an effect. Will give it a try! Thanks!
Thanks for the help man.
Another fun one I’ve got a preacher before offer recording my friend reading off from a self-help book and it was literally just like preaching in different voices. Every time I made a song complete a cappella no music something fun to play around with
That’s interesting! Will have to try that. Thanks!
I have read your comment 5 times and I still dont undersgand what your saying... Is thia cypher , is there 2 conversations going on at the same time? or is it a translation issue? I must know..What is the secret behind your comment,
Male/female voice is so random. Tried all strats and it just is a dice roll with the generate button until it happens
Yes - I’ve found the same thing. What helps a lot is using baritone, tenor, alto and soprano along with male or female in the song style area.
@@realmattdean If you want more male vocals then disliking your generated songs with female vocals will help tailor more songs being generated with male vocals. Also the entire prompt itself plays a part in which vocal is generated, meaning, just writing "male vocals" won't necessarily do it alone, especially if the prompt for example is "male vocals, pop, emotional, synth-pop", Suno will will 9 times out of 10 give you a female vocal based off of the "pop'' part regardless. Sometimes adding the exact genre you want last in the prompt can fix this issue. Most importantly though is going through and tailoring your profile with likes and dislikes on your songs. After a while you won't even need to worry about the prompt structure anymore and you'll still get the desired vocals 90% of the time. This works with specific voices as well. I've tailored my profile so much that almost all of my male vocals are pretty much the same voice now. Suno support team has also confirmed everything I've just said via emails I have with them.
This video is particularly interesting for the commentary on what Suno AI does not know how to do. At this time (Jan 2025) AI is excellent at some things but does not yet "get" certain concepts that humans manage routinely. I think that is the issue Matt is referencing at the beginning of this video. I truly love what AI is learning to provide for us, and I know it is in its "infancy". And, I appreciate the angst musicians, artists and other feel when they consider their career in the face of the looming AI juggernaut. However, Matt is giving valuable insight. The key to remaining relevant in this era of AI is to know the "ways of thinking" that are core to human behaviors. That means knowing music theory - which is the key to music complexity, novelty and interest - which is what will set human music apart from AI music (at least, that's how it seems to me). Thanks Matt.
Thanks John! So glad you enjoyed it and I'm glad I can be helpful. We're all just learning what these crazy things will do and it's all changing so fast. I think we'll need to rely on each other to find ways to use these to our advantage and not get left behind. Looking forward to hearing what you learn as we lean into the grand future that is being created around us!
would love suno to add when using the cover is that we can choose a persona and also that you can change lyrics, so you can do a cover in another language
that would be a great addition.
@schenier I agree, I've emailed Suno suggesting that to them just before they came out with personas.
Your Mister Roboto seems to be one way to get around Suno's ubiquitous pentameter (?) vocal DNA (architecture, family, sing-song-ie style). Thanx for the tips & tricks, esp. what doesn't work yet.
I hope that Suno will update to allow turning either an upload or a derivative of one to become a persona.
Thank you for your thoughts! It’s all growing soooo fast. It feels like by the time I get one video out there are changes but then it still won’t do male or female voices consistently…. 🤷 I’m glad this is helpful for you. Let me know what you find works for you!
If you want more consist vocalist for a song try: [VERSE 1: MALE VOCALIST], [PRE-CHORUS 1: MALE VOCALIST] , Etc. with Exclude Style: Female Vocalist. I've found this to work almost every time.. However, Suno is not always consist for the order of Male and Female vocalists for Duets. You will get a male and female vocalist but Suno may reverse the order of the vocalist or use which ever vocalist fits the lyric best. I've messed with this a lot... HTH
Thank you!
I agree about the inconsistency with Suno. It’s consistently inconsistent. It does seem to be a little better this week with male and female prompts than it was a few weeks ago so maybe it’s getting better. 🤷
Thank you. I will try it.
@@realmattdean it's likely due to the tailoring you're doing with likes and dislikes, as I said on a previous comment.
Thanks for the Knowledge👍👍👍👍👍
Glad to help!
I've been experimenting with the use of "," to separate elements inside [metatags]. Also, I've seen some kind of ":" use, EG [Intro: strings, Bulid-Up, In Crescendo]
Any luck? Would love to know what you find!
If you want, for example, female vocals. Exclude male vocals usually works.
Sounds good! Will give that a try. I’ve noticed this week that it’s been a little better at male or female prompts. Maybe it’s improving?
Thank you, Matt. Is there a place where I can provide a link to my song made on Suno AI for inviting suggestions on improvements? Thank you
Sure - you can put it in the comments. :)
Also very exciting is the possibility of finishing a song by Suno that I recorded instrumentally 10 years ago.
@@SteveM45 that’s amazing. I’ve been testing the upload feature before I make a video on it. But it seems really powerful. I’ve uploaded several instrumentals and it does and interesting job of adding different instruments or changing a guitar song into horns or piano…
@@realmattdean It no longer works in a useful way. It used to work fairly well, but both Suno and Udio have messed with it and it's no longer usable.
It doesn't work now. It just makes a garbled noise. It was working fine until they changed it recently.
@@group-music no
@@SteveM45 It's true I tell ya. 😎
I wish you would do the same videos for Udio... Great job in any case 👍🏻
Thank you! I’ve looked into it a little, but I’m actually trying to get a better handle on the ai video services. I’ve heard from some people that udio is really good in some areas. Will see what I can do!
مصاحبهای عمیق و تأثیرگذار! زندگی ایرج طالشی الهامبخش و پر از درسهای ارزشمند است. سپاس از این روایت زیبا. 🌟
Please provide that prompt which you mentioned in the comments section. So that will help the people those who are initial stage like me.
Sorry I missed that. I added some lyric prompts in the comments. Sometimes those get missed when I'm putting these up on youtube. If there is one specific one I missed, let me know and I'll add it. Thanks for watching!
I was using Suno and Udio to generate variations on my original music that I uploaded to their sites. This worked well until recently. Now, the results are completely unusable with very distorted random noises and many other bizarre issues. I wrote to them about this, but they just fobbed me off with some lame excuse. It used to work fine, but now it doesn't work at all. Maybe they have given up on that function which would be a shame. I don't want to make tracks that are just fully AI. I just want a virtual songwriting partner to provide possible variations on my own music.
I’m sorry to hear that! V4 seems to have a lot of issues. Sometimes it sounds very grating.
Great stuff. If I wanted to modulate up a chorus, how would I write the meta tags? [chorus] [modulate up] / [chorus: modulate up] / or just [modulate up] and Suno will recognize it as the chorus?
I’ve had some luck with [modulate up] but sometimes it ignores the prompt. I would do [modulate up] then [chorus] or [Bridge] [Modulate up] [Chorus]
Matt: great job, I write to you in English but I'm Latin, I make all my songs in Spanish, I think there's already a theme there, the whole page sees it in Spanish and with the original language (English) it's more specific and it does strange things to me. sometimes .
Regarding putting a series of chords and having the skynet obey it, I have obtained results like this: [G] + [D] +[F] . in most cases.
Have any singers stuttered? I have several, and it is almost impossible to fix it later, the replace works when it wants ahaha.
Problems with person: I made a profile and used it in a new song, at one point two verses from the original song appeared in the new song hahaha.
another with the extensions: I have an example of a one-minute part with a female singer and in the pair
new extensions a male singer appears.....
or else change the singers in each extension.
Sorry for being the messenger of crazy failures haha
Thank you Miguel
@@enifx wow - yes… Suno can be very very inconsistent. Sometimes when I do a persona, it does it as a male, then changes to female. It’s very strange that Suno brought back in the lyrics from an old song when you imported the persona!! Wow. 🤦♂️ let me know what else you find. I think I need to do a funny video next on all the hilarious errors that Suno does!!
Thanks man😁🫡
🕺
Man if Suno can create a full stems export of the song game changer
It does a voice and instrumental stem, but it’s pretty bad. It strips the song after creating it unlike a normal stem in protools and it’s full of a lot of bugs. I would love a version that could separate the instruments.
Yeah hopefully they are aware of the usefulness of that and add it
@@dagr8s815 try Moises, or if you really want the best stem separation possible get RipX Daw
Use another app for this purpose may helpful @@realmattdean
Interlude will also work. Instrumental interlude also
How can you get 2 different male voices or 2 different female voices as well as a group?
That’s a tricky one. I can get Suno to do a few harmonies and a few duets, but it’s pretty random. Call and respond works most of the time… ie:
[verse]
I love you (you’re the stars in the sky)
I love you (you make me fly)
Although time signature requests don’t work, if I randomly (e.g.) get a song in 6/8, I Extend from almost the very beginning and try different styles. Sometimes the results are good and gets me a 6/8 song.
Sounds like an interesting idea! Thank you!
Specifying [odd time signature] has worked for me. One in particular, I got a 7/8 verse and asked for [half time] on the chorus and got a half time 4/4. I also included "progressive" in my song description. It ended up fitting my lyrics so well I downloaded it then asked for [epic guitar solo] extending after the chorus lol. Took 4 tries but got something I was looking for and downloaded that. Took an intro I had created and crossfaded that into song I first downloaded then spliced the ending solo and turned out into a 7 minute Suno masterpiece.
can you do the mordern style dance/electronic phonk music!
I’m quite sure that is an option!
You may try brother 😮
When is the part 2 coming out
just now!!
for sound effects u need to be more simple like doing
[gunshot] or [scream] like very simple things so try like
[bang] or [spit] or something and for speaking voice I
do [talk] or [talking] and for female or male voice tend
to do [Mrs.Shauna] [Mr.Alex] wich seems to work more
accurate mostly, and if they duett u just say [duett
mrs.shauna and mr alex] but make sure to add duett to
the style promt, and for time just spellit out
Thanks! Will give it a try!
I recorded my voice and uploaded it to Suno. Thing is, I don't hear my cloned voice on a generated track, and I have repeated this many times, using my credits!
Can you briefly explain what I'm doing wrong?
Thank you!
I am working on a new video that will come out in a couple weeks regarding the upload and cover functions. I’ve had some really interesting results from instrumentals, but not from voices or from sample of horns. I’m not sure what’s going on with it yet, but I’ll report back as I discover what does and does not seem to work. I have found that the prompts used are VERY hit and miss with this feature - so I’m not surprised that it’s having trouble with voices.
@@realmattdean Okay, thanks. Will keep an eye out for that video!
@elnidogaming I’ve had pretty good success using my own voice, and while it’s not perfect, it sounds more like me compared to the Suno-generated voices. Here’s the process I follow:
1. Upload a sample of you briefly singing something with no background noise, just your voice.
2. Use that audio file to extend a new song.
3. Set the 'extend from' timing to where you want the song to start from your voice sample. If you don’t want the sample itself in your song, you can extend from a later point (e.g., 00:08). To create a gap before the song starts, add [break] in the lyrics box for the first few lines. For example:
Lyrics box:
[break]
[break]
[break]
Your voice won’t be imitated perfectly, but it’ll be close. You may need to go through several iterations to get the generated voice sounding more like yours.
Hope this helps!
@@KreoMelodyOfficial Thank you so much, I'll give this a go! 🙏
i asked 2 friends, if the voice sound like mine, and yes , it says around 30% . So yeah it kinda work
[Spoken] has about a 70% success rate for me personally
oh, and it could be confirmation bias but for say rock, putting 90s rock or 2007 etc seemed to help me skew towards what I was looking for
Also for voice expression I get a lot of utility out of prefixing voice e.g sultry female voice or soft southern accent etc. it seems to drastically shift the voice chosen.
That’s great to know! Thank you. I’m going to compile a list of what prompts others are getting to work so we can all contribute. Will try what you’re doing!
Hi Matt, I added you in Suno. I have also a very Long testing time in suno with different Commands and formating . You can here it in my songs. Very interessting are "Lied ist aus".
@@SteveM45 great! Would love to see what’s working for you. I’m
Hoping the more we point out what does and doesn’t work, the more Suno will make adjustments.
I have experimented with your Song "I Need you"😊
My Name = Sound4Soul
following. :)
suni cant tell difference between guy and girl voice... for me... i got crazy lyrics but the voices are jacking me up...
its so bizarre how difficult it is to recognize the most simple of things.
Your not 100% accurate on the voices being matched with genre, i have made many songs using other voices, the best way to control it is to use the lyrics section phonetically, most my lyrics are Meta tags and phonetics, it looks almost like code, this way I have had success controling everything from dj scratching to james brown-like voice to a drunk clown....
I’m quite sure I have a LOT to learn about getting this to work properly. That’s amazing that you’ve found some solutions. I’d love to see an example of what you’re talking about and I’m sure others would as well!
20:30 time Signature is in the Genre, e.g. waltz = 3/4
@@SteveM45 interesting. That seems to fall into the idea that Suno is just playing by ear rather than responding to specific directions. But it means that we can’t seem to make something like an odd rock beat in 7/8??
Just bare in mind that Suno owns any music you make using their app... if you use it in a RUclips video you will eventually get a copyright strike.... your "new band member" owns the band's work.
Not according to their legal documents. If you are using the paid service, you own the copyright. It’s only if you are using the free version that you do not. However, the copyright issue with ai music is a long way from being settled.
Of course you will get the strike, because suno protects your work to be stolen and uploaded by anybody as their creation.
The problem is there seems to be no white list solution yet to allow you to upload your own works to your own youtube account. And there is no other way to prove your rights than your suno account itself.
I think we need a digital watermark solution to sign digital works in the files itself to be able to prove ownership.
Sound effects works with *
really?? I haven't tried using anything other than [ ] and ( ) - does * help with any other prompts I wonder? Which sound fx have you found to work?
@realmattdean thunder, gun shots...
20:12 length 3:00 workt often for me.
@@SteveM45 interesting! I tried :30, :60, 1:30…. Nothing seemed to work. Most songs seem to naturally end between 2:30 and 3:30 so I didn’t try those. Thanks for letting us know what worked for you!
10:43 ahhh
😆7 min mark im out
🤷
Black voice wtf is that 😆
Suno is horrible. Udio all the way baby. So much better quality. In Suno in every song there is this noise in the background. Horrible.
It does have some issues. Will be looking into similar tutorials on Udio soon and will do a comparison.