To be honest, however, there has always been a large portion of music that was just a poor imitation of something that wasn’t very good in the first place, it just had money behind it to market it better. So does it really matter if a human wrote it? I guess the way I feel about it is that change is usually uncomfortable in some sense but there is no need to make it worse than it already is. Artists and musicians will always be innovating and expressing themselves in new and exciting ways and the paradigm has already been shifting toward live music anyway. I personally think it’s going to be a while still before AI powered robots are going to be able to put on a show that is as exciting as the Stooges, Parliament, Oh Sees, and thousands upon thousands of bands and performers playing shows every night all over the world. I think the main difference will be that people will write songs because they have that need to express themselves through music whether they get the big royalties or not and the live show will be thing that makes them money. As I said, it’s really been going that way anyhow.
So with that said I have no fear of AI taking my songwriting away from me. Quite the opposite. It’s really just another tool for me to use. That does not mean I just hit a button and collect a song written by AI and get any satisfaction from that. No, I still much prefer to write all my own music and perform the parts on some sort of instrument or synthesizer. That’s just something I love to do and the music I make this way is a thousand times more satisfying than working with AI, but there are times when having AI generate something for me to start with is very practical and even entertaining in a different way. At this point in my life I am willing to accept a little help in the composition department when I want something quick for fun or I have to create music in a style I am not particularly good at. Both of those criteria were met when I wanted to start making music for the video game I am writing, Space Smuggler Blooze.
So I started out with a few cloud based AI music generating platforms and got some entertaining results on occasion but nothing that impressive really. Then I found AIVA, a full featured digital audio workstation with AI music generation capabilities. This set of tools allowed me to really collaborate with AI instead of just getting a finished product out of a cookie cutter style generator. WIth AIVA I can take the midi data that the AI generates and customize it as much as I feel like, choosing the instruments, the effects, the mix, and I can even get in and tweak notes, tempos, song structures, you name it.
The great thing AIVA does is gives you a standalone DAW that you can install on your local system which plugs into their AI and sound processing API’s allowing you to make edits as you please down to very specific notes or phrases, choose from hundreds of nicely designed instruments including synth emulators for each individual track, mix levels from track to track, change tempo and arrange songs by sections of predetermined bar lengths, and also add effects from a fairly limited effects rack.
With the free plan you get midi downloads as well as mixed high quality mp3 for a maximum of three songs per month. This works for me as I am going to be compressing the video game music anyway and I can get the mix pretty good in the AIWA DAW. I simply create my composition, with AI throwing out a whole song or individual sections in one of a few hundred styles. I have been using 8bit and Commodore which create very old-school video game songs that I just beef up the instruments on and customize sections that don’t sound right to me for a nice hybrid video game instrumental.
They also have a $14/month, and a $40/month plan, the latter allowing you up to 500 downloads a month and stem downloads. While that would be nice, it really is overkill for anything I am going to be doing with AI. It’s not like I am trying to become an AI assisted songwriting machine. No, 3 songs a month is plenty for me as that is about all I ever really feel like doing with AI generating music creatively speaking. Strictly video game music. For my own music I would never use AI because it would defeat the purpose of writing an original song that is my own.
The next step once I have something I like that is mixed well in AIWA is I export the mp3 at 256kbps (pretty high quality) and bring it into my favorite DAW Ardour (By far the best open source DAW, or if you have ProTools, Ableton, or any full featured DAW you would use that) where I add any other instruments I might want to polish it with, add vocal track (for the video games I use a slew of online AI voice generators), and do the final mixing and mastering.
While making music with AI is not as deeply satisfying as writing and performing all the parts myself it still has some very useful applications such as making background music for a video, or in my case video game music where I can go from A to Z within a couple hours and have a decent entertaining song. When I do everything from scratch I will spend days on it. For me this is very useful.
That about sums it up for this months installment of Making Music With AI. Next I will go into some more specific details about how to use this workflow for anyone that is interested.