Electronic Music Automated Generation

Context

As of 2024, we have decided to become an AI studio, for pushing forward projects that intertwin AI technology with a very specific topic. In this new venture, we are focusing on electronic music generation. As part the GenAI movement, we dived into it by the pleasure that we have to solve the ever-lasting problem of creativity in AI.

We strongly believe that AI can foster the creative process and help more musicians to become producer of their own music, with help of AI that would alleviate the technical music production challenges !

We recommend the following article by Andreesen Horowitz, that explores in the bright way the future of the music Industy and how it is reshaped by AI : https://a16z.com/the-future-of-music-how-generative-ai-is-transforming-the-music-industry/

Partner’s needs

Automatic image evaluation of online marketing assets is important for various reasons

The startup in our AI studio needed an GenAI solution that would enable electronic music creation from a prompt, for music creators, music prosumers and companies accross industries, that need quality free-of-rights music for their showrooms, infomercials and workspaces or tiktok videos ! We wanted to build a solution that empowers everyone to create, twist, continue and deliver quality music right from a “prompt-to-music” interface.

Solutions Implemented

We created a state-of-the-art GenAI music generation algorithm, based on open-source materials that enable our wannabe artists create music.

The technology being proprietary, we cannot delve into details into the specifics of the code. Yet, our challenge was to generate free-of-rights electronic music : so we harvested free of rights database, even outside the electronic music scope.

Because we think that Beethoven, Mozart but not only can be remixed into a modern way and help boost the deep house music’s melody quality ! We chose to include them into the training data and also worked on a huge proprietary database of tags to qualify music upfront, in order to make the prompting more effective and more efficient.

Electronic Music Automated Generation

Results

In less than 1 month, we have built a Minimum Viable Algorithm that we could launch and test into real conditions, helping us in the process of testing product-market-fit as well as the AI ability to help real-life music producers to generate free-of-rights tracks of their own and deploy it to their application.