Secrets – March 2025
It is well known that music affects our emotions, our behavior or even the way we interact with each other. Different studies have explored music’s effects in various contexts, including its ability to accelerate patient recovery and reduce painkiller dependence in medical settings. However, generalizing these findings is challenging due to several limitations in the current music environment.
One major issue is the lack of standardized methods for collecting information on music’s effects, making it difficult to compare results across different studies and research teams. Additionally, studies on music’s non-monetary value often generate qualitative data, leading to unstructured textual documents that are hard to compare across regions, markets, application domains, or listener profiles.
So, how can we leverage the existing knowledge about music’s impact and provide analytical tools to communicate this value to various users and stakeholders in the music value chain? The answer lies in two main concepts: the creation of standardized repositories and the use of Large Language Models (LLMs).
Standardized Repositories, LLMs and Music Data: The Perfect Blend
Having a common repository where studies on music value can be stored and linked is the Rosetta stone for finding common insights and generating cross-cutting knowledge. The Music360 project aims to provide such a repository, offering an open platform to load and exchange information about music’s monetary and non-monetary value. However, counting with the technical support to store data is only part of the solution. The second necessary piece is to count with an adequate mechanism to connect quantitative and qualitative unstructured data in an efficient, affordable, and user-friendly way. It is in this second case that Large Language Models or LLMs enter the scene.
LLMs have gained significant attention for their ability to interpret and generate new knowledge from text documents, supported by companies like Microsoft with ChatGPT and Google with Gemini. With a music-oriented LLM it would be possible to analyze the textual data about music impact linked to specific musical information available in the Music360 repository. Thus, we can create a powerful tool for analyzing music’s impact in different contexts.
Imagine being able to query the music data and studies available to find the most effective type of music for patient treatment in specific medical fields or the preferred music for customers in retail stores. The possibilities are almost limitless, depending only on the constant enrichment provided by research teams on the value of music, and the need (or curiosity) of music users about what music can tell them.
This is a development in progress, and we expect the first LLM-based music value application to be available by the end of this year in the context of the Music360 Project. So, stay tuned to find out what your tunes say about you.
Music360 