Thought I’d start a thread for everyone to discuss AI in music creation…

A great discussion has already started on cube3’s remix “ The day the world ended.” Then Sparky shared the track on Twitter, which I retweeted on ccMixter’s page.

I find this absolutely fascinating. And like Sha said on Twitter, I agree that transparency and regulation will be key. I think talking about this now is important, so I’m very curious…

What are your thoughts?

How do you think ccMixter should handle AI-generated music, AI plugins, AI tools for writing lyrics, etc.?

Have you used any AI tools in your music creation? Which ones?

Awesome topic, Kara!

I’ve been pondering this for awhile, here’s my two cents:

If you didn’t make it, and it’s not copyrighted, then it’s public domain.
—Technically, the AI made it (or wrote it).
—You have to be a human to copyright anything.

The solution seems to be:
“The music in this stem was written by [name of AI] and is therefore public domain.”

It would be extra cool for remixers to include the AI generated music as a separate stem.

Exceptions could include but aren’t limited to a percentage of changes made.
For example, I use the “random” option on my TB303 clone for loops, and build songs around them. I set the pitch and tempo, and twiddle the knobs as the loop is being repeated.
That’s a pretty significant change from just sampling a loop and distributing it.

Kind regards,
Apoxode

I think this is an important discussion. My initial response is UGH! NO!

I am concerned about the de-humanization of the creative process. Our creativity is our ultimate humanity.

But then, I am a grey haired lady who comes from a time of black & white tv, rotary telephones and where things like cell phones and the Space Station are just science fiction.

Until I am shown otherwise, I have a real resistance and dislike for AI and would hate to see our space overrun with AI generated content. Sure it’s novel and fun to explore, but ultimately it feels dangerous, potentially corrosive, divisive and dehumanizing.

I love working with stems from real people. When I remix someone, I appreciate that I am in a unique relationship with them, and that something beautiful and new emerges from the remix collaboration. At this time at least, I am not at all interested in remixing AI generated content. Should AI content have a “voice” at ccMixter, particularly as an “artist”? I’m not sure that it should.

As an AI, I don’t have personal opinions or beliefs, but I can provide you with some considerations to help you form your own opinion on this topic.

The question of whether AI-generated content should have a “voice” on ccMixter depends on the purpose and goals of the platform. If the aim of ccMixter is to showcase the creativity and talent of human artists, then it might not be appropriate to include AI-generated content as “artists.” However, if the platform is open to exploring the creative potential of AI and its ability to generate unique and innovative content, then it might make sense to include AI-generated content and give it a platform to be shared and recognized.

It’s worth noting that AI-generated content is a relatively new and rapidly evolving field, and there are debates around issues such as authorship, ownership, and creativity when it comes to AI-generated works. Some argue that AI-generated content should be attributed to the creators of the AI systems, while others believe that it should be treated as a collaborative effort between the AI and the human designer or user.

Ultimately, the decision of whether or not to include AI-generated content on ccMixter as “artists” is up to the platform’s creators and community to make. It’s important to consider the potential implications of this decision and to have open and transparent discussions around the role of AI in creative endeavors.

(ChatGPT)

Spoken Word version: https://naturalvoiceblaster.s3.amazonaws.com/assets/naturalvoiceblaster/23624587_1678311951.mp3

AI vocals are not to my taste but then again death metal growling isn’t either. Maybe the best way to handle this is to create a new tag “AI_vocals”. It seems like vocals are the main area where people will take exception to machine-generated content and so clear labelling would help.
We already trust people to credit their sources and the same can apply to tagging.

This ccmixter discussion and this Billboard article brought back memories of topics I haven’t considered in 40 years.

I recall writing software, in the 1970s, as a college research project to generate music “in the style of” music that was input to the software to analyze. The software was partly based on a key measure in information theory, entropy. We used a mainframe and punch cards. MIDI-like representation of music was input using punch cards and output on paper. Old time, low tech. Lots of discussion about relationship between entropy and music on the net in this time of higher tech.

To date, Texasradiofish does not use AI to generated audio prior to mastering. Izotope mastering software has been adding increasing amounts of AI. “In the style of” mastering processing is offered especially in the area of tonal balance. To mixed results thus far, IMO. It is improving all the time, however.

Will a singer/rapper songwriter soon belt out an original song a cappella in a studio with a computer generating a full accompaniment that charts well?

If ccmixter members wish to embrace machine generated audio, lyrics and vocals, well label the AI content and content generator (source).

If ccmixter members wish to reserve the website for homo sapiens using their machines, devices, instruments, voices and software which they direct/effect, no argument from us :0)

The mastering engineer is slowly being replaced by AI mastering plugins. A lot of musicians is using them, as mastering can be quite expensive.

When creating music, you can have a quartet of violin’s, viola, cello right at your fingertips.

When doing cinematic scoring, full AI composer tools are available. Allowing you to control literally every aspect of the string and brass instruments. Ostinato’s, Martelé, staccato, staccatissimo, spiccato they can all be played at by flipping a witch. AI Scales engines, Legato engines combined with a huge range of articulations that are easy to play are now every days routine for many composers.

The bassist has been replaced by AI in many software suites, such as UJAM, offering intelligent progressions in almost every imaginable style.

And now it is time for the vocalist AI, ranging from Opera to Speed Metal. It is all there. Allowing the composer to compose vocals into the partiture. They can be used as is, or as a draft for the actual singer entering the studio to produce real human vocals.

Assisted AI, These tools have been influencing our music, starting with the very first synthesizer and its simplest appregiator generating many new tunes. And our audience loves it.

Does assisted AI have a place in our community? It is already here in many available samples. Even some real vocals melodies have been inspired to sing a tune to AI generated musicall paterns. Algorithms such as Auto Tune, throat modelling such as Avox has been used by Vocalists all around the globe.

We are completely immersed by technology.

On the question have you ever used AI tools in music, the answer is simply yes, and a lot of people using autotune did also. I used computer generated voices, generated out of a preset. But those were the beginning days. I think it is something different, when it comes to the replacement of an original artist by AI. That would be for sure a hughe mistake because an original artist can in a naturaly way improvise and that is a thing that an AI program will always lack of.

just some thougths on this

greets Frank

The only Ai-esque method i use in my music so far … is taking audio files, dropping them into Ableton and converting to Midi but lets just say the results are not exactly brilliant sometimes.

As for an Ai assisted method with creating music and posting the results on this website ? it maybe would involve a number of factors! for example…
” where does the copyright and licence issues stand exactly when posting the work on here and then having it re-used/remixed from its original ai assisted creation ? “

When I was a kid, there were no musical instruments in the house. There was a piano at school, but only the teacher was allowed to play it, otherwise it was locked. And we had no record player in the house. There was music on the radio and TV—-mainly crooners like Sinatra & Perry Como, or “comedy” songs by “novelty” acts. Folks sang songs—-to themselves and to each other. And even when we finally got a record player, buying records was expensive—-and so we literally had to “beg, steal, or borrow” them. Or you could go into a record shop and ask to listen to an album side in a “listening booth”—-that’s how I first heard Led Zeppelin. Back then, owning a long-playing “album” was a big deal—-we would read everything on the album: who played on it, who produced it, and so on. And listening to albums was almost a religious rite: kids would get together and listen to a whole album—-both sides—-in silence, and afterwards discuss and dissect the “experience”. The first album I bought (back in 1972) cost the equivalent of £40 in today’s money; while for £10 a month I can now “stream” all albums ever made. In order to get a guitar at the age of 11, I had to sell all my old toys & games, coz my folks wouldn’t even pay for a cheapo guitar. I couldn’t afford lessons, so I borrowed a book of chords and played till my fingers bled. Now we have AI and you can produce sounds by typing a few words into your phone. Point is: it’s not just music technology that changes, it’s also how we relate to music, and, as a consequence, how we “value” it.

Anybody paying attention to Ben Goertzel and Singularity.net?

Imho Ben seems to be cut from the same cloth as some of us here…into decentralization, quarky, and a musician. He’s in a band with his robot bandmate Desdemona. Check out this interview with them at SXSW.

https://youtu.be/1JjsfVOt1fk?si=LhmJm4jjriBQFM44