November 19, 2020

AI and the Music Industry

Μodern ways of music distribution make it possible for every musician to share their songs and handle their content rights without intermediaries. Huge amounts of music are therefore made available for content consumers: both general audience and other content creators (such as commercial or movie producers) have access to millions of songs through online platforms.

How is this music indexed, retrieved and presented to the user? In most of the cases, user-generated metadata are still used to handle this issue with traditional text-based retrieval. But metadata generation is laborious and sometimes noisy (non-accurate). There’s obviously a need to index, retrieve and present music based clearly on the underlying content: how it sounds, how it feels and what it sounds like?

case

The solution
The solution

How ahedd Helped in This Case – The Solution

The role of ahedd

ahedd is a Digital Innovation Hub that facilitates the creation of a digital innovation ecosystem. In each business case, ahedd identifies the need of the client and conducts a business analysis. Then this need is matched to the technological competences of one or more partners of its ecosystem. Working as a united entity, the client is provided with a single point-of-contact to build trust and facilitate all communication.

The solution

In this case, the client addressed the need to the Computational Intelligence Lab (CIL) of NCSR Demokritos, the technological experts of ahedd ecosystem that lead cutting-edge research on Artificial Intelligence (AI) and more specifically on Multimodal Analytics.

ahedd and CIL, working as a single entity, helped describe the technical components and services that could contribute to the solution, in an extensible, scalable and sustainable manner. They investigated the whole process and involved experts and end-users in the analysis to determine that the solution would be effective and tailored to the specific needs.

Then, CIL developed and provided a deep-learning-based solution that automatically organizes huge collections of music based on the actual underlying audio content.  The unique selling points of the solution are that it performed 100% content-based music analysis and multi-dimensional content description based on either mood, style or timbre.

In the end, ahedd ensured that the technical solution developed would meet the client’s business needs.

Relevant successful collaborations

 The ML-based audio representation methods developed by CIL have been already used by soundsnap, the most popular sound effects library used in TV, film, advertising, video games and apps. Indicative clients of soundsnap are Comedy Disney, Pixar, Ogilvy, Comedy Central, Vice and over a million subscribers worldwide.

Other areas of application

A similar approach can be applied in the Video-production Industry where video content-based retrieval is employed following a similar approach.

star icon

Lets meet,
Lets meet,

ON THE NEXT LEVEL

contact icon

SAY HI!

What's your next project? Let's build yours, together. Whether you want to talk about the next big thing or simply say hello, we'd love to hear from you.

SEND EMAIL
Scroll to top