Catalogue size
60,000+
‍
Products used
Tag & Search with Dashboard
Please note, this case study is no longer active.
Business Problem
Hipgnosis Song Fund is a Music IP Investment and Management company renowned for acquiring some of the best music catalogues in the world.
Hipgnosis acquires new catalogues at speed, and in combinations not previously seen in the industry before, so searching for the most appropriate placement could be a challenge when representing a new and exciting combination of tracks. Each incoming catalogue had previously been stored using different methods, so the task of finding and organizing the music became exponentially more difficult.
The main job of Music Funds such as Hipgnosis is to ensure that every single piece of music delivers a reliable Return on Investment. Ensuring that the music is tagged and searchable was the first step. Here’s how we helped:
The Solution Explained: Tag
The Hipgnosis team used our off-the-shelf tagging product.
Musiio's AI instantly tagged new acquisitions so that the team was able to search and find new tracks without adding, changing, or updating metadata manually. The tags included genres, moods, vocal gender and presence, an energy spectrum, key, BPM, and more.
This provided the team with a centralized system and unified taxonomy that they could learn and agree upon, which helped Hipgnosis to “clean up” messy metadata, and ensure that no mystery genres or tags appeared while the team was searching through the catalogue.
How does tagging work?
When Musiio's AI received audio input, it converted that file into a spectrogram - a graphical representation of various discrete qualities of audio. It then analyzed that spectrogram, comparing it to the 2 centuries of music it had been trained on, and returned a set of descriptive tags with >90% accuracy (see example below).
‍
The Solution Explained: Search
Hipgnosis used Musiio's no-code, drag-and-drop Search dashboard to tag, search, playlist, and send their music to clients for placement. New catalogues could be uploaded, processed, and added to their dashboard on the same day.
From there, they could use audio reference search (think reverse image search for audio) with YouTube links or client tracks, filter with tags, and send out playlists in response to sync and licensing briefs. Turning days-long searches into seconds-long searches.
With Musiio’s solutions, all of Hignosis’ tracks were just a few clicks away when needed, which maximized the ROI of every single track.
How did search work?
Similar to tagging, Musiio's AI converted audio files into spectrograms, which were used directly, and independent of metadata, to search the entire catalogue for the closest match based on up to 1,500 data points per track.
This was far more detailed than a metadata search (usually a maximum of 20 keywords per track). The AI picked up similarities in less verbally quantifiable musical characteristics: vocal tone, instrumental density, reverb type, and sound effects, for example.
Hipgnosis was able to use Musiio’s tags to narrow down an initial set of results. Perhaps a client had requested something similar to a reference track but wanted a slower version, or definitely a male vocal.
While Search worked independently of Tagging, both technologies could be combined to great effect. Setting a filter to focus on tracks tagged with Male-only vocals or a lower BPM allowed Hipgnosis to determine the closest match in just a few steps.
Not only does Musiio's A.I. answer a longstanding question, but it's ready to go - and it works. Crucially the team at Musiio have worked with us, to customise their platform to suit our exact requirements. This bespoke tool now allows us to instantaneously filter our 60,000 songs - in the ways that meet the needs of music supervisors, directors, advertising agencies, producers and broadcasters alike - at the press of a button."
Tom Stingemore, EVP Sync & Creative at Hipgnosis Songs.
Interested in how AI can help your business? Contact sales@musiio.com
