The Convergence of Music and AI: Enhancing Brand Identity through Personalized Playlists, a Collaboration between Brandtrack and Collective AI

Introduction

Brandtrack is a company specializing in creating musical atmospheres for commercial spaces, from bars and restaurants to spas, gyms, shopping malls, and luxury hotels. Their playlists are designed to enhance each client's brand identity, perfectly aligning with the atmosphere they seek to create, thereby enriching the experience of their consumers.

Passion for Music and Data

The key to Brandtrack's success lies in the integration of advanced artificial intelligence with human expertise, achieving a harmonious blend of passion for music and data science. Renowned companies like McDonald's, Hilton, and Santander trust Brandtrack not only for their musical expertise but also for their ability to offer a personalized sound experience powered by cutting-edge technology.

Advanced Technology

This state-of-the-art system, which enhances the work of music experts with data-driven tools, was developed by Collective AI. Our technology combines:

• An extensive database of musical metadata: It represents the music industry as a complex network of interconnected songs, albums, and artists, forming "musical communities" that facilitate the identification of new trends and the creation of more coherent and appealing playlists. To manage these relationships, we use Neo4j as our graph database, which allows us to leverage all the advanced algorithms of its Graph Data Science Library, optimizing the recommendation and organization of musical content.
• State-of-the-art AI models: They enable the automatic extraction of acoustic features to characterize musical tracks, as well as the analysis of song lyrics to obtain a contextual interpretation of their content. In cases where lyrics are not available on the web, we have developed a custom transcription pipeline that ensures the accurate retrieval of this data. This pipeline first identifies whether a track is vocal or instrumental using a specialized classifier; if it is vocal, it then separates the voice from the instruments using Meta's Demucs model before transcribing the lyrics with OpenAI's Whisper model.
• Graph algorithms: They generate personalized music recommendations based on customer preferences and music profiles, as well as playback history. These algorithms identify songs that are not only similar in style but also work well together to create the desired atmosphere.

Efficient Automation

Additionally, automated data pipelines, deployed as Step Functions in AWS, enable the efficient processing of tracks that Brandtrack adds to its catalog daily. With a catalog of over 70,000 carefully selected songs, these automated workflows make the collection and extraction of valuable information an agile and precise process, freeing up music curators from tedious tasks and allowing them to focus on what truly matters: creativity and musical innovation.

The collaboration between Brandtrack and Collective AI has resulted in significant improvements. Not only has the amount of available metadata been greatly expanded, but valuable information about song lyrics has also been incorporated, which was previously out of reach. By automating these processes, daily operations have been simplified, allowing music curators to optimize their time and focus on adding artistic value instead of performing manual tasks.

At Collective AI, we are experts in using artificial intelligence and automation to turn challenges into opportunities. If your company is looking to optimize its processes or leverage AI to achieve a real impact, we are here to help you take the next step toward innovation.