A few months back I was worked on my mixing skills, or lack of, to be more accurate. I wrote a song called "Mistake" that was more of a Female pop song. It was ok.

Recently I decided to mess around a bit more with AI and music, as many have. Some of these are in the same genre or style, but a slightly different feel. In all, for examples, this one goes to 11. So, it's the same chorus from the same song lyrics for all 11. I decided to call it "Many Mistakes." I would be curious any thoughts.

MANY Mistakes

After getting some input I thought about having a discussion around how we could use this. I know some feel that's been discussed to death, but I feel that discussion is just beginning. We may be in trouble.

thanks all! I've missed ya. A lot of life going on right now.

-Chad

Hi, Chad.

I think that most of the results could easily pass for well done pop music.

Heck, it took me a minute to notice when it got to the end the music and started playing the next song.

As to how we can use it, I'm not sure.

For one thing, no output generated by an AI qualifies for a copyright. That means that anyone could strip the vocals off your song, replace them with another vocal, and there's nothing you could do about it.

These programs have been trained by crawling through the internet and violating copyright, so ethically they're a mess. I think people who think these programs are "creative" fail to realize the massive amount of data that is used to train these programs. Programs like Midjourney process billions of images. Think of how many country songs these programs have been trained on in order to get flawless steel guitar licks.

As a songwriting tool, they're meant to replace the musicians and composers. They do too much to be an idea generator, because in the end, all you're doing is assembling a collage of ideas that were generated by the AI.

This is truly astonishing technology, but they're one-button solutions designed to replace composers, musicians, and mixers - not tools for songwriters.

Chad,

I enjoyed listening to the first two songs, will have a listen to the rest later.

As Dcuny says could easily pass for well done pop music, and adds

"I think people who think these programs are "creative" fail to realize the massive amount of data that is used to train these programs"

I suppose in a sense songwriters are trained by listening to the songs that have gone before, so in a sense no big deal in that regard.

The copyright issue is a big thing though I agree.

I think what it will boil down to eventually is the general public will not care one iota, if its a good song, whether AI wrote it or a songwriter, and neither would I.

To be honest I just love this Technology, and I think the traditional way of putting a song together can happily co-exist with the AI way.

Anyway though biab is not AI (though at times it may seem it is employing some sort of basic AI) I think we shouldn't be complaining as users of biab that modern AI is going beyond our comfort zone.

Sorry to say it (again), but as much as I love technology, I find, and will always find THIS technology really, deeply sad.

I kinda totally love this. I'd love to see you turn out a CD with every track done to the latest time standard for songs. (Is it still EXACTLY 3:34?) 10 tracks of different iterations of the same song titled "Mistake".

I liken this long running AI debate to the time way back when I caught all kids of grief because I liked techno music so much. All I heard was that sequenced parts controlled by sequencing software and played back through a series of MIDI controlled instruments was "cheating". Everybody I knew hated it. May I list a few bands they "hated"?

Howard Jones
The Thompson Twins
Talk Talk
Gary Neuman
The Human League
Haircut 100
Flock Of Seagulls
Duran Duran
Daft Punk
Pet Shop Boys
Tears For Fears
A-Ha
Depeche Mode
Simple Minds
Eurythmics

What makes me laugh is knowing that a lot of the people who put those bands down went on to do solo acts with, you guessed it, backing tracks. And in a true stroke of irony, many of them use backing tracks that they bought rather than programming them. Or even perhaps (gasp) writing their own songs!! The reason they do it is, of course, because they can keep all the money with no band members to pay. And to push the irony further, many of them look down at karaoke. Um.....

Karaoke, correctly written kara oke, is Japanese for "empty orchestra", which is exactly what they are doing. Music with no musicians performing it. The first karaoke machine was called The Sparko Box, a little cube that flashed flights as it played, and the guy who invented it, Shigeichi Negishi, died last month at 100 years of age.

And here we are with software created by the very much alive Dr. Gannon. And discussing AI tools that can replace it.

Irony can be so ironic sometimes.

“The rule should never be such that human creators stand to gain more from repeatedly clicking a button to generate massive amounts of AI-produced materials than from putting their hearts, souls, experiences, skills, talents, and emotions into expressive works of art.”

https://www.musicbusinessworldwide.com/nmpa-generative-ai-is-the-greatest-risk-to-the-human-creative-class-that-has-ever-existed/

I am sorry to say, but it all sounds crap to me.

I need to hear music created by real musicians that sing with their own voices and play their own instruments and let their soul and spirit out in songs that are created.

It's a dozen examples, an original and 11 variations. How to use the idea is how BIAB has done it since its inception. My BIAB version has 11,165 styles that the process you used can be applied using the StyleMaker to create 11,165 versions of the Chorus lyrics from this posting. One can use Spleeter (BIAB has the capability to include Spleeter or a similar product embedded in later versions of BIAB) to isolate the vocals from your post and apply them into 11,165 versions with very similar results to this process.

More common is to use the process in BIAB to create complete, original, and unique copyrightable songs with arrangements that can be shared and published.

The 11,165 Styles can each be user modified, modified or merged into per-song styles of creating new styles to add to the 11,165. The RealTrack page lists 4,700+ instruments plus some styles include midi patches and SuperMidi patches. The higher versions of BIAB have more than 5,000 hours of studio recorded audio. Open the SGU Chord Sheet and it defaults to two sub-Styles. The program allows for each Chord Sheet to have up to 24 sub-styles that have the feature to modify a Style on a per song basis. Drum patterns, instruments can be muted or replaced with other instruments.

Regarding RealTracks, the BIAB 24 Track Mixer has the feature of each track having an eleven track sub-mixer that can place the original plus 10 other configurable RealTrack instruments on each track. 24 tracks times 11 instruments provide 254 instruments playing possibilities in user-programable configurations per song.

BIAB isn't cheating or ironic by any stretch of the imagination because it requires human intelligence at every step to manipulate it into a song.

I seriously doubt whether, if presented with several "real music" samples interspersed with several AI samples like Chad presented, you could consistently pick which is which! And even assuming you could do it, it won't be long before you won't be able to.

I am sorry to say it all does not sound like crap to ME...
I need to hear real honest music, by real players too, but this, to me is quite surprising, and maybe a little disturbing, because it's doesn't sound like crap... ... to me...

I have not followed much of the discussions, but find the subject fascinating...
I'd love to know how this production was created... what "ai" software? ai assisted?
Yes, trouble ahead...

I gotta say, reluctantly, I'm kinda with Eddie on this!

It has always struck me as hilariously ironic that folks who use software like BIAB to create much/most of their music, get all hot and bothered when some new technology like AI comes along and offers ways to do the same thing a bit faster with even less involvement!

Heck, we've even had long arguments here where folks who replace live band members with BIAB bass, drums and rhythm guitar parts in their "live" performances, then turn around and complain that open mic participants are somehow stealing their gig opportunities!

If you're using BIAB or similar tools to create your music you're not much different from the kid using AI to produce his music. You just drew the line in a different place.

Now, as for my personal preference, I would never knowingly go see a performance where the band is using backing tracks OR AI in their "live" performance. And I'd prefer not to knowingly listen to AI music. But going forward I don't think I'll have much choice.

Hi chad
That’s interesting. What was your workflow and what AI tools did you use? Look forward to learning more.

J3 IKR?? When I saw that I almost fell out of my chair!


Even more jarring when the whole musical component, to include solos, are presented as a "band". And when the performer is playing something non-descript like bass... wow. Is there a bass solo in every song there, Jaco, so you can show off your chops? Who pays to see a nobody play BASS?


Similar in concept to the early days of 7 foot basketball players. If they were on the opposing team they were "oversized freaks who are ruining the integrity of the game", but when they are on YOUR team they are "a remarkable athlete who takes great advantage of his size".

J3, remember these are the same ilk that said TV was just a fad, decried the invention of the automobile, and refuse to use anything that wasn't the norm when they were young. They also still pay bills with checks that they send in the mail. (Man, it's 2024!!!) I am an old bas***d (I included the ***terisks) and I absolutely LOVE all of this great technology. I can sit in a Discord server and talk to people all over the world in real time. I was chatting with some scambaiting friends last night. We had England, Ireland, Denmark, Australia, New Zealand, Canada and the US represented. We can all do baiting calls together, which we do on Wednesday and Thursday. We play characters (with voice changers) and sometimes occupy a scammer for 3 hours straight, during which they can't be calling YOU. This week I played my gangster character, and at one point the guy in England was the bank customer service agent and a woman in Texas was my secretary. We collect bank account and remote connection information and pass it along to the right people to harm the scammer effort as much as we can.

I am trying to imagine my father, who died 34 years ago at 73, understanding what was going on and what kind of voodoo allowed it to happen. Much like the people in the audience in 1993 when Todd Rundgren toured with behind his No World Order album and his "band" was 8 racks of MIDI controlled gear arranged in a circle around him playing lush, lavish arrangements while he pranced around the stage and sang. A woman sitting behind me complained the whole night because he wasn't playing "Hello It's Me". (I REALLY wanted to turn around and say "Lady, that was 20 years ago. People grow. Try it.") I thought it was great. She had no idea how anything worked and had NO intention of embracing anything "different".

Sometimes you just have to turn up the squelch and eliminate the noise.

I had a lot of fun listening to this. I thought songs 5 & 10 are the funniest because they both sound like showbiz dram or something you'd hear in a musical. The modern country pretty much hit the mark and the gospel was spot on. What got me the most though is how earnest all the vocals sounded.

There's a huge difference between these AI programs and BiaB, sample libraries, and MIDI tracks: the AI companies don't have the rights to use the voices and instruments that are in their songs.

They make of point of not telling where they get their materials from, but here's a project in 2020 that described how it gathered its source material (emphasis added):


See: https://openai.com/research/jukebox

That is, these programs trawl through millions of copyrighted songs to get their training material. No artist has granted them rights to use their materials, and no artist is compensated.

The companies creating the AI programs do a number of things to make it difficult to determine where the source materials come from.

One of the most obvious is omitting artist names in tags. So although the AI is capable of rendering a song with the voice/instrument/arrangement of a particular artist, there's no way to request the AI do so.

And since a specific voice can't be requested, you'll end up getting a voice that's a combines the attributes of similar voices - enough so that the original singer can't be identified.

However, here's something Udio produced when prompted to generate something in the style of the Beatles:

https://twitter.com/i/status/1778900566917165512

Udio didn't pay the rights holders of The Beatles songs to use their songs, and Paul McCartney didn't authorize Udio to use his voice.

In my mind, this is theft (well, technically massive copyright infringement), disguised the same way that money laundering hides the source of illegal profits.

There's a big difference between using BIAB (or any other piece of software) as a tool to assist the musician in the process of music creation and using a computer program to completely replace the musician AND the music creation process.

I personally don't find these songs sound bad at all (actually they sound pretty professional to me), I just find they sound pretty generic, uninteresting and soulless, it's just plastic music, like 90% of commercial music nowadays. So far so good.

But asking a computer program to compose a song about, say, falling in love, pressing a button, and getting a complete and finished product has absolutely nothing to do with the beautiful, complex and ancient art of music composition. And the same can be said about asking an AI to write a novel, a poem, or a movie script for you. There is and there will never be absolutely no value in such a thing.

What I find deeply sad is that some people seems to start to find this type of pseudo "art" interesting and acceptable. If this tendency goes on, who would want to invest time and effort in learning to play an instrument in the future? Who will want to spend years in learning to speak a language? Who will want to read a book? Who will want to spend years in an art school? Who is going to want to make any effort to learn anything if everyone thinks that computers do everything better and more efficiently?

If this tendency goes on, we will be much more stupid in a few years than we are today. Thank you, but count me out.

You make very reasonable points. Cerio. Thank you for that.

I think I like what AI might have to offer. So far I have not got involved with it, as I'm concerned I might become reliant, perhaps too reliant.

I think I can see where it can be unbelievably beneficial (advances in medicine, cures for disease, science, technology), but as far as art-forms exist, the human mind needs to always nurture that creativity. Venture as we dare to, but keep in mind the real reasons that art exists. Because of us.

Regardless, I certainly don't mean to derail this thread. The O/P has generously shared their own results which no doubt consist of informative, interesting and enlightening (perhaps even somewhat surprising) results. The results are incredibly amazing. Imagine thinking that a microprocessor could do that a decade ago (or less)? No? I can't either.
(I just hope Skynet is not monitoring )

Intriguing post. I'm not a fan of the IDEA of AI songs, even when they sound as much like modern popular music as these do. On the other hand, I'd LOVE to have AI work up my music videos. Is that weird?

That's pretty much how I felt when I first heard some of the examples as well.

I feel there could be many uses for it. I'll say a bit more further down in my reply.


Very true. I can see where some people might absolutely hate that. I'm not entirely sure where I fall on that yet either. I'm still trying to figure out where that line is as far as at what point is actually something you could copywrite. I've read some things that were somewhat vague. I'm wondering if that's on purpose? Any insights on this? I know you're researched it A LOT, which is greatly appreciated.


This aspect of it has me go back and forth a lot. I feel I know your stance on it from what you've posted. I think we would disagree; and I'm ok with that. I know you are a great guy, who puts a lot of thought into what you are saying, and it comes from a place of truly researching this subject...you're educated. We've just arrived at different conclusions and I completely respect yours.


This is truly astonishing technology, but they're one-button solutions designed to replace composers, musicians, and mixers - not tools for songwriters.

As of now, I would say it's definitely not there yet. Having worked on these projects I learned a lot...especially the limitations and annoyance of having to figure out how to communicate those ideas into something useable. Even at that, after the fact, I still had to do a fair amount of editing to get to somewhat work. Sure, the sound might be mostly right, but the structure, and uniqueness of the song have a long way to go. I do think the will definitely be closer to if not, exactly what you are saying down the road though. I'm sure you already know that though.

Thanks. I had fun doing them...also frustration though.


Totally agree. Something will need to be figured out. Obviously not the same, but the whole "sampling" thing that went down had some similar issues that had to be sorted out. But those were directly using the artists work as it is.

.

I tend to be more in this camp as well. If I can hear great music or a great song, that's really what matters to me. I do understand where people would have issue though as well.


Completely agree. I'm excited for certain aspects of it, and wonder how certain questionable things will be addressed.


As others have pointed out, these tools also are not true AI either. I personally believe that it's similar in concept to what people have been trying to accomplish with BIAB. Not in every case with every person obviously, but (just my take) if you can't play an instrument, or can't play it well enough that you are giving a program instructions on how to do it, then possibly generating multiple takes to get what you want...at the heart of it it seems similar not matter the method or program. I hope that made sense?

I appreciate the reply

No need to apologize. I'm just curious what about it makes you sad? There's so many different ways someone could mean that. I am truly wanting to understand what you meant.

Thanks!

Ok, there's a lot there! ha! I do know where you're coming from. I completely enjoy the bands you mentioned! I actually saw Depeche Mode in concert for the first time in 1990. The music was fantastic and the lights were all programmed to visually go with all the little production things they do. I was so geeked out it was pathetic! I loved it! Btw, talk about making your songs interesting! They were masters.

The band that opened for them one of the times was Nitzer Ebb. If you haven't heard of them, they were a great lead up to DM.

Did you listen to Information Society at all by chance? A bit more happy pop, but still did some cool things.

I'm not sure we have the right to tell others that any more than they should be able to dictate our art, in my opinion.

If someone makes something that I don't care for, I still respect it, but it's not for me. I'm not really all that sure that you're sorry for your comment; or maybe would have taken the time to word it better instead of pre-apologizing. Just a thought.

One last thing, I couldn't find any songs you posted. Keep at it. I truly mean that. Ask questions, get feedback, and improve as you go. I have a feeling you may be way more critical of your own music than you are mine. Don't let that stop you. We are all here to help. We all suck sometimes, and we all hit a gem sometimes.

I'm not sure how completely original they are. Not challenging you, I'm truly saying I don't know. I know when I do regenerations SOME of the licks, for let's say a guitar, turn out having some of the same parts. It's nice though, because you also get variations to work with and build, say, a solo. As an extremely weak guitar player, I like these options!


What I posted also took manipulating to get it usable. MANY takes did not work out well. To fully construct a song using this technology would be somewhat difficult. Not impossible, but a fair amount of work. I'm just finding people seem to have all sort of opinions on how much it's too much when it comes to having computers helping in the creation of a song.

You are TOTALLY correct in the amount of options we are given with BIAB. It is and always will be a wonder product to me. You stated all of that way better than I did!

Always enjoy your comments my friend. Everything I've said was meant with total respect.

Depending on the system it's played on, it might be difficult at times already. I completely agree that the time of not being able to distinguish is not that far off. I'm not sure how long, but given how fast it's progressing...pretty crazy.

Thanks for the response!

That was my exact experience. It was a bit jaw dropping at first as I tried to figure out what was AI and what wasn't. That in itself spoke volumes to me.

It was a combination of a few things.

I used my own lyrics that I had written. No AI at all.

I used a site called Udio.com (made an edit for spelling) to create most of what you head musically. I took A LOT of regenerations as it does get many things wrong in song structure and also delivering the lyrics with random words at times. Other times, it just would generate additional lyrics if I didn't have enough in a section. I wouldn't use those and instead would rewrite the lyrics.

In an actual song, I use a DAW to have to assemble multiple takes and design the song structure. It someone just starts vocals wherever at time and jumps into a chorus at a weird time. THEN, for the second verse, as an example, it doesn't use the melody from the first verse but instead generates an entirely new melody. That makes it REALLY difficult. Also, there are no repeating of any number of choruses. Thus, have to use the DAW to copy and past the chorus.
I am actually big on shortening parts of the chorus or expanding it to add unexpected interest. I'm not really seeing this as possible...yet.

So for those who think this is a click a button one time kind of thing; consider educating yourself. It's so not there. I do believe in time it will get there. I would think it would be much easier than what they've done up to this point.

Hope that helps. Ask away on questions. David is more the expert though.

I look at it very similarly. Some of the people who argue this is too far are the very people that musicians who don't use BIAB look at in that same way. It's kind of like the speed you drive at. Whatever it is is the right speed. People who don't drive your speed are driving too fast or too slow. MY speed is the RIGHT speed.

Hi Joanne! I was hoping you would comment because of your interest along these lines. I love that attitude!

I did answer what I used and how above. The technology still has things that need improvement before it gets to the point of writing songs for us before we even ask it to.

On a personal note, I love what you've been doing and working on. Congrats on the success with it. I could see it pairing well with some of these technologies! (As I think BIAB already recognized!)

I'm glad you did. Yes the two Metal Showtunes were fun. One was supposed to be a bit more like a choir...almost like a Tran Siberian goes to church kind of thing. The second one I was going for Metal ABBA a bit.

Since I can't sing, the vocals are what most impressed me.

Appreciate the comments my friend.

First, you know I respect you BIG TIME, I hope. I think you are making fantastic points all the way around. I may not agree with some of them now, but that doesn't mean I disagree either. I more of the mind of "I don't know" enough about some of it to really say with confidence on some of it.

As an example, and there are many I believe, BIAB has a "style" for Dire Straights that sounds pretty spot on to me. I think if you generated it, most people who know the band would identify it as being that. I'm fairly certain PG Music didn't pay Dire Straights for that signature sound. I don't know though. IF they didn't, I'm still ok with it. It's a sound. A signature sound though. I see this as being a similar situation. Just my opinion my friend. Still see your points.


I agree, and I know this point has been made. To me, it's similar to what what we do as humans in many ways. We take reference materials to learn what we need to do then spend the rest of our lives spitting our those things in new combinations. I think the sticking point many times I've seen from people is when WE do it, it's creativity. When a program does it, it's WRONG.

Pablo Picasso is often credited assaying “good artists borrow, great artists steal.” Some people say he wasn't the first to say it. To me, this would be ironic if he took that from someone else.


Isn't that kind of saying that it's a new voice then. If the voice sounds similar but isn't, it simply isn't that voice, in my opinion.


I am SO curious how they did this. Specifically what prompts they used. I find it EXTEREMLY difficult to get it to sound similar to some of the voices that I was hoping I could get it to sound like for my own purposes of fun. I've got a bit of a voice crush on Halsey and have not come even REMOTELY close to that sound. Like, at all. I must be missing something.


I think you said this in a really effective way. You always have such good points. I sort of feel like they shouldn't have to pay them, in my opinion because we aren't generating any song that sounds exactly like any Beatles song that exists. Yes, it sure sounds like them...but that song doesn't exist anywhere else. I feel the same could be said for the voice issue.

Your points are truly fantastic and I can't tell you how much I appreciate them because they make me think, and re-evaluate my thoughts on the subject. Thanks for that! We may not agree but I'm not here to argue. I'm HearToLearn.

Chad, thanks for that explanation... did you mean to type "Edio" or Udio?
I played around with Udio and was impressed with the website, and the tech, but I don't know what I'd ever use it for... I must say, many of the song examples were high quality, as were those that you produced... After editing in your DAW, were you able to publish any of your finished songs back onto the original site? Were you able to download any of your creations as WAV files? Any luck generating videos on their site?
Thanks again Chad

Dang! Good catch. Yes I meant Udio!

After editing in the DAW, no I was not able to publish them back up to the site. I haven't tried it though either. I just get the feeling you probably can't.

I didn't download any of the examples or songs I played around with. I just captured them in my DAW.

I haven't tried the video aspect. I honestly didn't even know it was a thing. So, I'll have to look into it. Thanks for bringing it up!

Overall I am hoping at some point it would be able to have more control over the song structures...and maybe export multi tracks? Am I asking too much?

Thanks for the reply!

David, I can understand why you would say that they used Beatle songs and Paul McCartney's voice because Udio was able to get something musical that sounded a lot like a Beatle song as sung by Paul McCartney. But, it was not a Beatle song and may not have been sung by Paul McCartney.

Sound-A-Like music has been a thing in country and rock music for as long as I can remember listening to music. Heck, the first "Beatles" album my family had was not by the Beatles but by a group of session musicians and singers that replicated songs from the Beatles first album before the album was available in the US.
A Dutch group, +++ Stars On 45 +++ performed amazing replications of not just Beatle songs but also songs like "Funkytown", "Boogie Nights", the Andrew Sisters, A Star Wars medley, ABBA and more.

The Wikipedia article about the making of the recordings is fascinating reading especially since it goes into an overview of how this was all reconstructed from scratch and digital technology like sampling was not available. +++ 45 RPM Medley +++ is a short 4:44 video of the 7" 45 RPM record. The original release was an extended play 12" recording that was 9:44 in length.

To me the funniest idea is they covered The Archies song "Sugar, Sugar". The Archies were themselves session musicians and singers brought together to sings songs for a Saturday morning cartoon show based on comic book characters.

The point I'm making is not that your wrong as I don't think your far off the mark. Only is it likely that original and copycat recordings were used as models. The copycats should add their own character to the end result making the end result different from what it might be with all originals.