A neural network may help an online music service to spot songs with the potential to go big.

What’s new: Musiio uses AI to identify specific attributes and qualities in recorded music. Online audio distributor SoundCloud purchased the Singapore-based startup, which was valued at $10 million last year, for an undisclosed sum.

How it works: Musiio trained its model on a proprietary database of songs, each tagged with dozens of labels including genre, vocalist’s gender, instruments featured, and emotions expressed.

  • Musiio’s technology drives a number of services including automated tagging of up to 1 million songs a day, audio search, a tool that combs a publisher’s catalog for weak material, and one that helps agents discover new talent.
  • A demo released in 2019 enabled users to upload a song and generate labels for genre, key, tempo, energy level, and emotion. For instance, the demo might have labeled a song as instrumental, “moderate energy” with “small variance,” and a 72 percent probability of being “dark.”

Behind the news: A number of companies offer AI-powered tools designed to enable recording companies, artists, and fans to squeeze more value out of music.

  • Fwaygo lets artists upload short video clips, which an algorithm will recommend based on a listener’s preferences. Fwaygo recently partnered with music distributor TuneCore, which supplies music to Amazon, iTunes, and Spotify.
  • InGrooves, a music marketing firm owned by Universal, patented a system that generates social media posts that feature songs selected by an algorithm to appeal to a certain audience.

Why it matters: Millions of new songs are released every year. Amid the deluge, AI can help distributors recognize potential hits, recording companies identify talent, fans find music they like, and musicians create sounds that stand out. Of course, the makings of a hit include social dynamics among listeners — presumably that’s where acquirer SoundCloud comes in.

We’re thinking: According to models, this edition of The Batch has moderate energy with high variance and a 72 percent chance of being powerful.


Subscribe to The Batch

Stay updated with weekly AI News and Insights delivered to your inbox