Streamers have high expectations from the music apps they use in 2020. A well curated, personalised service that learns musical taste and makes daily recommendations is a baseline criteria for selecting a streaming app in 2020. In the future, the stickier services will invest in features that are able to;

  • Give users more control over their recommendations
  • Improve the active search/self-curation experience

Most streaming services have 50 million+ tracks and add 40,000 tracks per day to their platforms, so the problem isn’t that there isn’t enough content, it’s that the technology that is used to recommend it, needs improvement.

What we do know is that musical tastes differ from person to person and from one week to the next, which is why using lookalike audiences and popularity scores can lead to poor, or seemingly obvious echo-chamber-like recommendations. 

With Musiio’s Playlist API, recommendations are generated by ‘listening’ to the anatomical makeup of the audio of the last 100 songs a user has listened to. With this data, we are able to make accurate audio-based suggestions that every listener will love. 

In order to test the AI and the approach, we recently conducted a blind listening experiment to test our AI's ability to select a great playlist of tracks against 5 of the best streaming services in the world. Here is one result:

This graph shows results for perceived accuracy of recommendations based on a particular seed-track. This test was conducted blindly by a group of non-Musiio listeners. See all data in the Full Report.

To access the full report, please email us at

Share this story