It's a new era for music production, and artificial intelligence is leading the charge. Just last week, it was announced that Apple acquired a London-based company called AI Music. According to AI Music's website, the goal of their software is to give consumers the power to choose the music they want, edited to fit their needs or create dynamic solutions that adapt to fit their audiences.
Whether we like it or not, AI music generators are becoming more and more sophisticated. They're able to create melodies that sound remarkably like those written by human composers. As this technology continues to evolve, it's only a matter of time before AI-written music becomes indistinguishable from traditional music composition. So what does this mean for the future of the music industry?
The first paragraph of this article was written by me. The second article was written by Jasper, an artificial intelligence copywriting tool. I thought it would be interesting to see what GPT-3 would say about the future of AI music, given a neutral prompt. It looks like, at least in my experiment, AI is self-confident and optimistic about its own future. We'll see whether that actually pans out.
For the remainder of the article, I'll share my opinion (not the robot's) on the current state of AI music apps. These tools allow you to generate MIDI ideas using an AI model trained on thousands of human compositions. The results can sound surprisingly good!
Defining AI Music
By definition, AI music is a composition that's been generated by machine learning algorithms. It's a new form of music created without human intervention or input. These AI models have been trained on thousands of songs from various genres, giving them the ability to create melodies in styles ranging from classical to pop and everything in between.
How well AI music reproduces the characteristics of traditional music composition is still up for debate. But as AI technology continues to evolve, it's only a matter of time before AI music becomes indistinguishable from human-written compositions.
The Best AI Music Apps for Music Creatives
Artificial intelligence is changing the way we make and listen to music. AI music apps are a great way to explore this new technology, and they might even help you create a masterpiece of your own!
AIVA Song Generator
AIVA is an AI-powered music generator that works in your internet browser, using deep learning algorithms trained on over 30,000 human compositions. The ai model has been trained for years to produce full arrangements in almost any style of music. You don't need to play an instrument or know anything about making music to use it. Just select your parameters, like the genre of music and emotion that you're going for. AIVA does the rest.
Once you find something you like, you can export the entire track as MP3, MIDI, or WAV stems. Music producers can reference the files as a starting point for their own music. Just be aware that the free plan is limited to three downloads per month. So while you can drill down pretty deep and customize things like key signature and tempo, you'll probably need a paid plan to make the most of the app.
AudioCipher MIDI Plugin & Melody Generator
AudioCipher is a simple but powerful AI music app for music producers. It uses a transcoding algorithm to turn words into MIDI melodies. Instead of creating melodies based on genres of music or emotions, the app turns each letter of your phrase into a music note. Type words into the app, pick a key signature and AudioCipher will spit out MIDI melodies right within your DAW.
Try AudioCipher with the 100% free OpenAI Musenet to leverage neural networks in your word-to-MIDI adventure. Once your cipher is arranged and mixed down to a wav file, you can use OpenAI Jukebox to transform it into a full arrangement. The same way that Dalle 2 and MidJourney turn words into images, you will now be able to turn entire phrases into a song.
The premise of AudioCipher is that every word and thought carries a unique energetic signature. Producers can type in anything that comes to mind and use the MIDI output to compose new musical ideas in that style. Instead of a neural network dictating the melody, AudioCipher provides a MIDI note series. This keep the rhythm and cadence open to your own interpretation.
Use the rhythm randomization feature to create new melodic phrases with each MIDI file. This will change the note durations at random, to give a more human feel to your melodies, while still keeping a strong melody at the core of it all. Or you can use this MIDI plugin as a starting point for new ideas, completely reworking the note sequence into something new and unique.
If you're a producer looking for a fun way to get inspired on a regular basis, AudioCipher is the perfect tool for that. As a MIDI plugin, it loads right within your DAW as a VST/AU component. You can generate unlimited melodic variations on a single phrase or try out different words until you arrive at something that works.
Amper AI Music Generator
Amper is another powerful, browser-based music generator app. It delivers a similar feature set to AIVA, but with even more controls over the final output. Amper generates full songs based on the genre, mood and duration that you request. But more importantly, you get access to a dashboard where you can control the instruments that are kept in the mix. This way the audio files you export only include the content that you want.
You won't get MIDI files from Amper, which means you can't change as many details about the melody or virtual instrument. If you're looking for a similar type of app that generates AI music for free and exports to MIDI, we'll cover that next.
MuseNet MIDI Generator
MuseNet is one more browser-based app that generates music with artificial intelligence. But where AIVA and Amper focus on audio output only, MuseNet specializes in MIDI output. You can still control a few inputs like style, instruments, and length. But you'll need to select a musical theme from a list, so the algorithm has something to build upon. If you've already composed a MIDI idea, upload it directly to MuseNet and the app will elaborate on your song for you.
Play the MIDI output back right within your browser. If you like what you hear, you have the option to export it as an audio file or as MIDI. The benefit of exporting MIDI is that you can change the notes in your digital audio workstation.
One of the limitations with MuseNet is that they have a fairly limited collection of styles to choose from. Fortunately, with the option to seed your music, you'll always have ability to shape the kind of audio that comes out. As a workflow tool, it would be easier to have an app like this available right within the DAW. Hopefully in the future that option will become available!
Boomy AI Music Community
Honorable mention goes to Boomy, an AI music generator company that provides creators a path toward publication. Resembling AIVA and Amper in its feature suite, the app's biggest differentiator is the option to distribute your music on streaming platforms like Spotify, TikTok, Youtube, and more. At the time of this post, Boomy users have reportedly created 5,690,684 songs, the equivalent to around 5.95% of the world's human music.
The community aspect of Boomy is one of its greatest appeals. Users' tracks can be explored on the site and on their Discord, where roughly 1,000 users are currently registered. Whether the music the app creates is actually any good... that's for you to decide.
Excited by all these apps and looking for more alternatives still? Two other AI music apps that we enjoyed were SoundDraw and Jukebox. We wrote a full description of Jukebox in this article on AI song generators.
The Future of AI Music Apps
So with all these AI music composition tools available, are humans still going to be a part of making music in the future? I certainly hope so. AI music apps may get better and better over time, eventually able to produce masterful pieces of music that rival a human being's capabilities. There's some evidence for that already.
We could have a future where AI generated lyrics and music are prompted by humans, who make adjustments rather than composing from scratch.
We already have a lot of AI-created music in our world. The difference is that it's usually hidden behind layers of production. A producer or engineer has taken a piece of AI music generated by an algorithm and then tweaked it until it sounds "perfect." But what if we could remove those layers and hear the AI music as it was originally created?
I think there's a lot of potential in this idea. We will have to see if AI can create something new and different that we've never heard before. This could also be a chance to hear what a machine's mind sounds like. How does a computer think? Maybe if we listen closely enough, we'll find out.
Of course, a lot of the AI music that is created today unpleasant to our ears. But it's getting better all the time, and with advancements in deep learning and neural networks, we're sure to see some amazing things in the near future. Stay tuned!