Royalty-free music for commercial projects
User-generated content tagging & acoustic search
Exhaustive & consistent tagging and better UX
Jamendo is a music website and an open community of independent artists and music lovers. Jamendo offers the world's largest selection of royalty-free music. With a community of millions of members, the catalog comprises 500,000 tracks by 40,000 artists from 150 different countries.
The Jamendo Licensing platform (B2B) gives back millions of dollars to Artists, by offering royalty-free music for commercial projects (film, TV, commercials, apps...) as well as affordable in-store music solutions.
Clients include major brands such as Zara, Nespresso, Ikea, Mcdonald's, Four Seasons, Dior, TBWA, Ford, H&M.
Martin Guerber is the lead Music & Content Manager and worked before in record labels, concerts/events production and artist management.
Jamendo ‘s catalog is based on user generated content and offers the artists the opportunity to publish their own music. The Jamendo Licensing service acts as an intermediary between artists and third parties who wish to use the music in their projects (videos or in-store music). In order to connect artists with brands & corporates, Jamendo had to find a way to improve the discoverability of their music tracks and better curate their music.
Jamendo’s website used to only rely on semantic search technology based on the metadata attached to uploaded tracks. As the tracks are uploaded directly by the artists, the tagging is often either inconsistent, partial or even totally absent in some cases. Jamendo was looking for a solution to unlock the potential of their non indexed tracks and add value for both artists and their end clients. If people can't find music efficiently for their needs, your music catalog is losing business opportunities. Reviewing all the tracks by humans and keeping up-to-date with the incoming contents was an enormous task. This was not a solution.
Jamendo tried to increase the tagging process using crowdsourced online solutions. The results were disappointing and the quality criteria was not met. This is where Niland comes into the picture.
Every time new tracks are uploaded to Jamendo’s website, they are sent to Niland API. Using advanced techniques of machine learning and signal processing, Niland’s technology enriches every track with effective descriptive tags (genre, mood, instrument, tempo…). Niland’s technology provides a deep understanding of music and computes the acoustic similarity between tracks as a human would do.
Tags and similarities between tracks are then returned to Jamendo’s system to power the search and recommendation engines. Jamendo chose to work with an offline integration of the Niland API. Every day they push the new tracks to the API. The next day they can pull the data (tags and similar tracks) and store them in their own database. For text search they built a solR search engine. The solR configuration uses the tags and the associated relevance score : this way, when a user types “Happy”, the happiest tracks appear first in the results. The integration has been running for a year with no scalability issue since then.
Jamendo implemented Niland as a core tech for their platform relaunch in april 2015. Niland’s solution helped Jamendo to improve their User Experience. The new search engine is now fed with consistent tagging and delivers relevant results for Jamendo’s clients. Niland’s audio technology was essential to bring new features such as similar tracks and albums. It helps clients quickly find alternative tracks for their project and brings them an easier way to navigate through Jamendo’s content.
Niland’s solution is also used on a daily basis by the internal team of music supervisors. When receiving a brief like this :
Jamendo’s music supervisors use Niland’s dashboard to perform audio searches and find tracks that fit the references. The team also built more and better playlists using niland's search features.
Jamendo’s business is growing rapidly and will soon launch new features for the music listeners side of their website. Niland's technology will be used to recommend more and more personalized content dynamically through new discovery features. “Moreover, whenever we have problems or new features needed, we simply reach out to Niland’s impressive support team,” said Martin.
Product release, tech meet-ups...be the first to know.