Spread the love

Artificial Intelligence is

increasingly being used to generate all sorts of creative content like images, text and now even music! While AI generated music is not at the level of a professional composer yet, the tools have come a long way and can now produce surprisingly melodic and enjoyable tunes. In this post, let's take a look at some of the best AI tools available for generating original music compositions.

Anthropic Grammy

Anthropic is an AI safety startup that has created some really unique tools for AI assisted music creation called Grammy. Grammy is an AI assistant trained on a huge dataset of musical scores, lyrics and recordings to understand the complex relationships between different musical elements. Here are some of the cool features it offers:

  • Composition aid: You can describe the mood, style, tempo, instrumentation etc for the music you want Grammy to generate and it will compose an original piece matching your request. The AI understands tonality, melody, harmony, rhythm quite well to produce coherent multi-instrumental compositions.
  • Lyric generation: If you provide Grammy with a melody, it can analyse the rhythm and suggest lyrical phrases that fit the melody. This allows non-lyricists to have AI generated lyrics for their melodies.
  • Arrangement help: You can upload your MIDI files or sheet music and Grammy will suggest arrangements, introduce countermelodies, basslines, and refine the song structure based on music theory rules.
  • Remixing: It can take existing audio files and remix them by generating new melodies, rhythms, or transforming the mood based on parameters you set.
  • Interactive mode: Grammy also has an interactive canvas where you can sketch out melodies or change chords in real-time and the AI will respond musically to develop the composition further.

Overall Grammy has some very sophisticated music generation capabilities while maintaining an empathetic and helpful conversational interface. It is still in early access but shows a lot of promise as a creative AI tool for both novice and professional musicians.

Aiva Technologies – Aiva Studio

Aiva is another AI startup working on generative music technology. Their tool called Aiva Studio allows users to describe the mood, genre, instrumentation for the music and the AI will compose an original multi-track song based on that. Here are some key features:

  • Natural Language Composition: You describe the music verbally like "I want an uplifting piano ballad" and Aiva understands it to generate the song.
  • Real-time Collaboration: The AI keeps learning from your feedback on the generated music in real-time, making adjustments like changing the melody, adding counterpoints etc.
  • Different mediums: It can generate music as MIDI files, sheet music or Audio recordings for various instruments.
  • Variations Engine: Aiva Studio has a variations engine that can spawn new musical ideas derived from the original composition to help take the song in new directions.
  • Published Works: Some of Aiva's generated music has been published and performed by artists already showing the quality it can produce.

Overall Aiva Studio provides an easy to use interface for non-musicians to generate professionally sounding original music compositions with the AI handling all the technical music theory aspects behind the scenes. It has a very natural interactive mode of working with the AI.

Jukebox – OpenAI

OpenAI is well known for developing powerful generative AI models like GPT-3 and DALL-E. They have also released an AI system called Jukebox which was trained on a massive dataset of musical recordings spanning several genres, decades and languages. Here are some notable things it can do:

  • Song Generation: You can provide Jukebox with a description of the song you want like genre, mood, lyrics and it will compose and generate a full song as sheet music or audio.
  • Style Transfer: It allows taking existing songs and transferring their style like changing a pop song to sound like jazz or altering the tempo, instrumentation etc.
  • Fill-in-the-blanks: Provide Jukebox with an incomplete song like just the melody or drums part and it can intelligently complete the other parts to make a coherent whole song.
  • Remixing: Jukebox has the ability to take existing songs and remix them by generating new verses, choruses, solos or completely rearranging the song structure.
  • Converting formats: You can feed it audio recordings which it can transcribe into sheet music notation or vice versa.

While access to Jukebox is limited since it is research oriented, it shows the amazing possibilities of AI in automating complex musical tasks like composition, remixing with just natural language descriptions.

Anthropic Music 2.0

Anthropic has also released an updated version of their music AI tool called Music 2.0 which expands upon the capabilities of Grammy in many interesting ways:

  • Robust Music Understanding: It has a much stronger understanding of music concepts like timbre, rhythm, melody, harmony, tonality compared to previous versions through its self-supervised learning.
  • Better Composition: It can now generate multi-track compositions (drums, bass, lead melodies etc) across various genres matching provided styles seamlessly.
  • Lead Generation: Provides tools to generate catchy lead melodies, riffs, solos that can be added over the backing tracks composed by the AI.
  • Lyrics Assistant: Helps lyricists find rhyming words, suggest phrases fitting the melody, refine lyrics based on sentiment analysis.
  • Sound Design: Capable of timbre design for virtual instruments, effects like reverb, distortion based on the desired sonic properties.
  • Version Control: An interactive versioning system to keep multiple edits, variations of the compositions in parallel.
  • Distribution Features: Exports songs as sheet music, flattened audio tracks, STEM files for mixing directly in DAWs.

Music 2.0 shows Anthropic's continuing effort towards building powerful and production-ready tools for AI-assisted composition and arrangement across genres. Its natural language interface makes the whole music generation process quite seamless.

Jukedeck – Music Generation Platform

Jukedeck is another commercially focused generative music platform for creating original tracks. Here are some of its notable features:

  • Genre Selection: Pick from predefined genres like ambient, pop, jazz, movie trailers etc to filter the AI's output.
  • Mood Selectors: Sliders for attributes like energy, loudness, emotion help steer the AI towards the ideal mood.
  • Instrumentation: Choose the instrumentation like piano solo, full band, strings section etc.
  • Length Control: Set the duration of the generated track between 15secs to 5mins.
  • Refinement Tool: A refining AI suggests variations of the composition based on feedback.
  • Automatic Sectioning: Tracks are coherently arranged with introduction, verses, choruses etc.
  • Multi-track export: Export stems for mixing individual elements or full mixed track.

Jukedeck's web based drag-n-drop interface makes creating AI tracks very fast and easy. Their AI is trained on a unique copyright-free dataset enabling commercial usage of generated songs. Very handy for quick mood-setting backgrounds

Mid Journey – AI Art Music Creation

Midjourney is a popular AI image generation tool but it also allows creation of original musical pieces based on text prompts. Here are some interesting things it can do:

  • Song descriptions: Provide details of a desired song and it will generate the sheet music or scored audio.
  • Mood/Genre influenced: Influencing factors like dark, optimistic, jazz, EDM in prompts helps steer genres.
  • Specific instrument focus: Prompts centering on piano ballads, guitar solos, orchestral epics etc work well.
  • Stylistic variations: Asking for "remixed", "reimagined" versions of popular songs spawn new ideas.
  • Multitrack composition: Some prompts generate complete scored compositions with different instrumental layers.
  • Music + Art combinations: Prompts fusing music styles with visual elements like "Impressionist piano painting" generate sheet music + artwork pairs!

While the music generation is relatively basic, Midjourney shows how powerful text-to-music generation could become with continued development. The multimodal art+music responses are quite unique capabilities.

Anthropic Jukebox – On-device AI Music Generation

Anthropic has released another music AI tool called Jukebox which is optimised to run entirely on mobile/desktop devices without internet. Some key benefits of this local model are:

  • No Internet Required: Once downloaded, it works offline on phones/computers enabling music making anywhere.
  • Fast Deployment: The self-contained WebAssembly packages are very light and instantly load without server backends.
  • Privacy Preserving: Since it runs locally without sending data to servers, it doesn't collect or share any user data.
  • Unlimited Usage: No subscription limits as the AI model resides on device for on-demand access anytime.
  • Accessible Interface: Browser/app UIs provide natural-language songwriting workflows optimised for mobile.
  • Distribution Friendly: Exports AI generated songs as standard audio/MIDI files for sharing directly.

This on-device music AI approach lowers the technical barriers for widespread accessibility of generative music tools. It empowers independent music making anywhere without online dependencies.

IBM Watson Beat

  • Natural Language Interface: Just describe the kind of beat/music you want in plain English and Watson Beat will generate an original track based on your description.
  • Focus on Beats & Loops: Unlike tools focused on full songs, Watson Beat specialises in generating individual instrumental beats, rhythmic loops and slices that can be mixed together.
  • Wide Genre Support: It understands a variety of music genres like hip-hop, EDM, pop, jazz, classical Indian etc. and can produce beats within those styles.
  • Customization Options: You can further tweak attributes like BPM, duration, instrumentation of the generated beats based on your mixing/production needs.
  • Remixing Capabilities: Watson Beat has a Remix option that takes existing vocals/samples and generates new beats to complement them automatically.
  • Interactive Refining: The AI keeps learning from your real-time feedback on the generated beats, tweaking elements like snare patterns, hi-hat variations to improve it iteratively.
  • Multi-track Beat Stems: Exports the full beat as individual audio stems for each drum element (kick, snare etc.), melodic layers that can be mixed flexibly.
  • Web & Desktop Apps: It has a web dashboard for online access along with desktop apps for Mac, Windows for easier studio integration.


Q1. Is the music completely generated by the AI or do humans provide some input?

A1. While the AI systems compose the music automatically, most tools require some human input like describing the desired mood, genre, instrumentation for the music. This ensures the output tracks match human preferences. Some advanced models can generate fully original songs from just a title or description as well.

Q2. How different is AI generated music from human-composed music?

A2. AI music is generally quite good for basic compositions but lacks the complexity and nuances of professional human music. AI focuses more on melody, harmony and catching popular music patterns vs subtle aesthetic/emotional qualities. But tools continue advancing towards human-level mastery.

Q3. Can AI tools replace human musicians/composers?

A3. Not yet – while AI excels at repetitive patterns, it struggles with creativity, improvisation and expressiveness central to mastering complex instruments. Most experts believe AI will augment rather than replace human musicians by automating basic composition/arrangement work.

Q4. Is the AI-generated music open for commercial use?

A4. This depends on the individual tool. Some only allow non-commercial use while others like Jukedeck are specifically designed for commercial music licences. It's best to check each platform's usage policy before relying on AI music for commercial projects.

Q5. Is the quality consistent across different genres of music?

A5. While most tools support multiple genres, their abilities vary – some excel at pop/EDM genres with strong repetitive patterns but classical or jazz involving complex improvisation remain challenging. Quality also depends on the amount of relevant training data for each genre.

Q6. How private is personal data/music shared with AI tools?

A6. Reputable tools do not share or sell users' music data and take privacy seriously. However, basic details like genre/style preferences may be used to improve the model for all. Anthropic's local on-device models collect no data.


In conclusion, artificial intelligence and machine learning techniques have come a remarkably long way in synthesising original and enjoyable music with just a few natural language instructions. While AI music generation is still developing and cannot match the creativity of skilled human artists, these tools have huge potential for aiding composition, remixing, and arrangement tasks to boost productivity.

The field is progressing at a rapid pace with new models capable of more complex musical storytelling across genres as they learn from ever larger datasets. On-device and non-internet dependent AI applications also help address privacy and accessibility barriers for widespread adoption. With continued enhancement of music theoretical understanding, Expressiveness, stylistic variations and human-like improvisation, AI will surely get more integrated into mainstream music creation workflows.

Tools covered here like Grammy, Aiva Studio, Jukebox, Anthropic Music 2.0, Jukedeck are great options for both amateur and professional music producers to kickstart songwriting or augment existing works. Meanwhile, Arts+AI platforms like Midjourney indicate how music generation could be fused with visual/narrative arts in multi-modal forms. Going forward, conversational assistive systems may even become creative partners for composing original soundtracks.

In all, AI is undeniably transforming how music is conceptualised, produced and consumed with each passing year. While human creativity remains unrivalled, these technologies empower new artistic frontiers by automating mundane production tasks to allow focusing more on the artistic spirit. With ethical safeguards, we may see AI augment human potential across many creative fields in the future. The journey of AI and music has only just begun.