Elevating Music Discovery with AI/ML

June 7, 2022

Today we're featuring a guest blog from Cory Nakabayashi who leads our AI/ML Strategy and has been involved since our founding:

Artificial intelligence and machine learning play a significant role in unlocking our strategy of continuous content

At majr, we start with experiments that test and push against our assumptions.  Today, we’re actively experimenting with novel ways of creating and returning value back to creators and their music.  These experiments are expansive and varied, but each have a profound impact in changing the ways we unlock new signal pathways by leveraging AI/ML to rethink the experience that artists and fans expect.

For example, we've taken progressive strides towards expanding the digital representations of songs and albums by putting them in the context of their evolution over time.  In other words, embedding information that describes the growth and change of a song over the music-writing process.  In a world where song information is measured in a handful of text fields, static strings like title and genre, a comprehensive representation of musical evolution opens up exciting new possibilities in both analytics and discovery. 

For artists, this means next-generation tooling able to help fuel your own creative processes.  Imagine an ecosystem of tools that delivers intelligent songwriting suggestions or generates and pushes inspiration directly to you, all driven by and reasoned over with an advanced machine learning pipeline trained on hundreds of thousands of creatives that came before you.  Understanding the underlying signal in how songs are made, and how they evolve from first ideas into final cuts, also means extending the way we think about and deliver on the tools that support your creative process.

For fans, this means unlocking new ways to interact with the creative content of the artists you love.  We know that the cultural value of music will always exist beyond the play button and shallow metrics like streams.  It lives in the lifeblood of the fan communities that follow and support musicians, in roadtrip mixes and listening parties, and is stamped on the emotional roadmap of our lives.  By enriching creative content on majr, we are able to increasingly bring this comprehensive music experience to the digital world, powered by content graphs and digital tools that simply aren't available on traditional platforms. We think this is the key to delivering rabbithole experiences that fluidly allow you to explore the past, present and future of the music you love.

We believe that by applying machine learning models to understand how real songs have evolved may unlock insight into composition as a craft, leading to better artist tooling, more immersive fan experiences, and ultimately to providing more value to both fans and artists alike.

A hard look at the broken economics of the digital music experience has also forced us to reexamine the relationship models that drive valuable discovery.  Streaming companies like Spotify and Apple Music have hinged on plays as their primary lever of value and invested almost exclusively on pushing fan-to-artist discovery, leaving the other vectors, namely, fan-to-fan and artist-to-artist, underserved and under-monetized.  

By rehydrating the entire network of fans and artists and opening up the exchange of value in all directions, majr hopes to achieve two things: (i) the explosion of data and opportunity to inject ml-driven discovery and analytics models into this exchange, and (ii) the use of which will facilitate greater throughput on delivered value from all nodes of the network.  

What this looks like in the real world is a discovery and recommendation experience that is truly connective.  An enriched content representation allows our machine learning models access to richer signal information for connecting artists with their fan communities faster and more effectively, mitigating the "Spotify payola effect" by directly delivering content to the real communities that love them without worrying about the outdated algorithmic obstacles that "streaming-only" recommendation services are built upon.  We're also working on a new set of tools that build artist-to-artist connectivity to inspire a new creative of ecosystem of collaboration, and a fan-to-fan community-based approach to growing cultural communities on top of the content that they love, all powered by AI investments and experiments rooted in redelivering the true value of music back to you. 

At majr, we really do believe in the power of music, in our own lives and in the world at large, and are working hard to build a better place for music in the digital age.  

Artists deserve fair pay
This is the future we're building.
Supporting Artists