A Complete Guide to AI Music Apps in 2023 (All Categories)
Over the past three years, AI music apps have been springing up everywhere. It's an exciting time for those of us who enjoy tinkering with emerging technology. If you follow artificial intelligence closely, you've probably seen the viral social media posts that list out the latest AI software that you have to try.
Many of these aggregators include audio and music apps, but they tend to focus on AI loop generators and celebrity voice impersonation. In some cases they will cite tools that no longer work, but present them as if they were brand new.
Our goal here on the AudioCipher blog is to give a voice to the smaller AI music tools that the public is largely unaware of. There will inevitably be tools that don't make the list, either because we overlooked them or felt that they didn't meet a certain threshold of quality to warrant mentioning.
We'll start with some of the familiar entry-level tools that anyone can use, before venturing into software that appeals primarily to musicians.
Table of contents: Categories of AI music app
The first technical conversations about artificial intelligence and music are often traced to a 1974 computer music conference at Michigan State University. For those who want to explore the history of this field, journalist and Dolby machine learning expert Jordi Pons has penned a solid summary of AI music history.
Our list of AI music apps below were tested and found to be active as of May 2023. You can click on section titles or individual software names to skip down to that part of the page.
Section 1: AI music apps for beginners and non-musicians
Section 2: Text to Music AI software
Riffusion
Section 3: AI DAWs for musicians
Section 4: AI audio-to-midi converters
Section 5: Additional AI music app categories
AI music apps for beginners and non-musicians
To get started, we'll share a complete list of the most popular AI music apps that anyone can use to create a clean loop or song. By clean, I mean that it lacks the weird audio artifacts that you'll hear from neural audio synthesis software like Harmonai's Dance Diffusion models.
SOUNDRAW.io
Website: SOUNDRAW
Popularity: High
Difficulty level: Low
Generative Attributes: Mood, genre, theme, length, tempo, instruments
Music genres: Acoustic, Hip Hop, Beats, Funk, Pop, Drum n Bass, Trap, Tokyo night pop, Rock, Latin, House, Tropical House, Ambient, Orchestra, Electro & Dance, Electronica, Techno & Trance, RnB & Soul, Lofi Hip Hop
Cost: $19.99/month or $203.88/year
Users can explore existing tracks in the SOUNDRAW music database or hit create to start generating music from the available attributes like mood, genre and theme.
As of Mid-2023, the #aimusic hashtag on Twitter and TikTok feature songs created by Soundraw more prominently than other AI music app competitors.
A simplified AI music app, called Ecrett, identifies itself as Soundraw's "little brother" and claims to have generated over 54 million combinations of music.
Boomy AI Music Community
Website: Boomy
Popularity: High
Difficulty level: Low
Music Genres: Electronic, Rap beats, Lofi, Global Groove, Relaxing Meditation, Experimental
Generative Attributes: Genre, Instruments, Drums, Mixing, Sound Effects, Tempo
Cost: Three plans available
Free Plan: limited saves (undisclosed qty) and 5 album/single releases
Premium: ($2.99/month) Unlimited saves, downloads, 20 releases
Pro: ($9.99/month) Unlimited songs, releases, and priority rights requests
Boomy is an AI music generation community that lets users create original material and collect royalties through streaming platforms. The web app offers an extremely low barrier to entry, with just a few clicks required to create a song.
Over 14.5 million songs have been created with Boomy, representing 14% of the world's total recorded music. The app came under fire in early May 2023 when Spotify accused Boomy's creators of spamming the service with mobile farms that run up the streaming count.
Tens of thousands of songs were taken down and many sites incorrectly reported that these takedowns were due to Boomy's low quality music. Spotify clarified its position about the stream farms and reinstated the Boomy publishing pipeline a week later.
Soundful AI Music Creation
Website: Soundful
Popularity: Medium
Difficulty level: Low
Music Genres: Lofi hiphop, Hype beats, Relaxation, Club beats, Trap, Dance, Drill, Afro pop, Disco, Pop funk, Synth pop, EDM, Techno, RnB, Latin, Reggaeton
Generative Attributes: Genre, Instruments, Drums, Mixing, Sound Effects, Tempo
Cost: Three plans available
Free Plan: 10 downloads, 1 stem pack
Premium: ($9.99/month or $89/year) Unlimited downloads, 5 stem packs
Pro: Cost is undisclosed, custom solution, monetize your Soundful music
Soundful is more nuanced than Boomy in its approach to music generation. You can pick from a greater number of genres and music attributes.
During my experiments with Soundful, a "dark" beat in a major key produced a track in the relative minor key. This could lead beginners to believe a song is in "C major" when it is actually in "A minor". It's a bit reductive to think that a dark song has to be in a minor key. Meek Mill's song Splash Warning features a major pentatonic melody with a haunting pad under it, for example.
Shutterstock's Amper AI Music collection
Amper was another powerful, browser-based music generator app. It delivered a similar feature set to AIVA, but with more controls over the final output. For better or worse, Shutterstock acquired Amper in 2020 and assimilated the app's musical output into a static music library. As a result, the AI music app has been reduced to a stock audio service competing with other royalty free music sites like AudioJungle, Envato Music, Epidemic Sound, and Artlist.
MuseNet MIDI Generator
MuseNet was a free browser-based app from OpenAI that generated MIDI files using artificial intelligence. When it launched in 2019, users could select from a few parameters like style, instruments, and length. The MIDI would playback in your browser and there was an option to export the MIDI file for free.
An independent developer, Steven Waterman, built a UI layer on top of the MuseNet API called MuseTree. His design was based on branching nodes that allowed users to spin up a variations rapidly, and with more controls than the MuseNet website.
Unfortunately, OpenAI switched off the MuseNet API near the end of December 2022, the same week that ChatGPT came out. The Musenet page now features a static demo that mimics the MIDI generation functionality. It always generates the same MIDI output. MuseTree no longer works, due to the MuseNet API being decommissioned.
Text-to-Music software
Text-to-image apps like Dalle 2 and Midjourney have created a demand for similar experiences in music and video generation. As of 2023, there are a handful of AI music apps that offer this capability, with varying degrees of success.
AudioCipher: Text-to-MIDI VST
Website: AudioCipher
Cost: $29.99 one-time purchase (free lifetime version upgrades)
AI Music App Format: VST/AU plugin formats, Standalone option
Music output format: Midi
Target audience: Musicians who use a DAW
When we launched the AudioCipher text-to-midi plugin in 2020, we weren't intending to become an AI music app. We were expanding on a simpler, classic technique called the musical cryptogram.
In its current form, AudioCipher users control key signature, chord extensions, and rhythm automation. Additional randomization features for rhythm and chord construction deliver endless variations for your text. Musicians retain full agency over the instruments they choose and shape those chords and melodies into music. This creative process can later be shared with fans as a musical easter egg.
As AI text-to-music APIs become more accessible, we are exploring the possibility of upgrading our algorithm. Existing text-to-MIDI models have not demonstrated the quality and flexibility that we're after. Our product team is in talks with several ML engineers currently. We may end up training our own text-to-MIDI model if we can afford to. Revenue from app sales has funded research and ongoing app development to keep this effort alive.
Version 4.0 of AudioCipher is due for release in Fall 2023. Customers receive free upgrades whenever a new version comes out.
Mubert Text-to-audio web app
Website: Mubert
Cost: Licensing structure ranges from $19-499 per track for a 45 second track
AI Music App Format: Web app
Music output format: MP3 and Wav
Target audience: Musicians and non-musicians
Mubert is currently the only text-to-music software to generate full songs without unwanted sound artifacts. Their system intakes written descriptions and passes it through an ML model trained on a labeled audio dataset. Existing human-made loops in the database are modified using composition algorithms and a layer of sound design. This confers a distinct advantage to other text-to-audio services like Riffusion and Google's MusicLM app. It lacks the unwanted sonic artifacts produced during neural audio synthesis.
Riffusion
Website: Riffusion
Cost: Free
AI Music App Format: Web app
Music output format: MP3
Target audience: Musicians and non-musicians
Riffusion was published in December 2022, preceding Google's MusicLM app by nearly six months. The free web app uses Stable Diffusion's text-to-image generation to render spectrograms (images of sound) from user input. These spectrograms are then sonified (made audible) to produce music. It's a unique technique that produces short-length, low-fidelity clips. The audio lacks in quality but makes up for it in charm and entertainment value.
Google MusicLM
