Request for Extended API Access for AI-Based Spotify Thesis Project

[HrissTsvetanova]

Hello,

I am currently developing an AI-powered music assistant as part of my university thesis project, focused on enhancing the way users interact with Spotify through intelligent analysis, recommendations, and conversational interaction.

At this stage, my system is already functional and operates using the data currently available through Spotify’s API (primarily liked/saved tracks). Based on this data, I have implemented the following capabilities:

• Storage and structuring of liked tracks, including artist, album, release year, and metadata
• Automatic genre inference (especially for Drum & Bass subgenres such as neurofunk, jungle, liquid, techstep, etc.)
• Mood detection based on audio features (energy, valence, danceability, acousticness, etc.)
• Classification of tracks into mood categories such as: aggressive, melodic, atmospheric, energetic, uplifting, dark, etc.
• Sorting and organizing tracks by genre, mood, and historical period
• Identification of most frequent (top) artists based on listening data
• Basic recommendation engine based on:
• genre (e.g. "recommend neurofunk tracks")
• mood (e.g. "I want something aggressive")
• combined mood + genre queries

Additionally, I have implemented a natural-language query system that supports prompts such as:

"show all my Spotify tracks sorted by genre"

"sort all tracks in my Spotify liked songs by genre"

"recommend tracks"

"suggest jungle tracks"

"recommend me drum and bass classic"

"recommend me drum and bass only"

"I want something aggressive"

"I want something melodic"

"I'm in the mood for dark Drum"

"suggest something atmospheric"

"recommend emotional liquid"

"I want energetic jungle"

"which artist appears most in my liked songs"

"what Drum and Bass genres exist"

"when did jungle appear"

The system also supports historical genre analysis and era detection, allowing users to explore the origins and evolution of specific styles.

The next stage of my project (currently conceptual and partially designed) includes:

• A voice-based AI assistant (similar to Alexa), for example:

"Hey Spotify, recommend me tracks for chilling with friends at home."

"Hey Spotify, give me music for a late-night party."

"Hey Spotify, what did I listen to most this week?"

• Advanced mood and atmosphere detection based on listening patterns and context
• Weekly and monthly listening recaps (beyond the current annual Spotify recap)
• Discovery of new artists based on deep preference analysis
• Context-aware playlist generation (e.g. social events, studying, nightlife, etc.)
• More advanced recommendation logic using full catalog data
• Potential future extension: intelligent mixing / track sequencing (for advanced usage)

However, in order to fully realize these ideas, I would need broader access to Spotify data.

At the moment, my system is limited to what Spotify currently allows through the API (mainly liked songs). While this is sufficient for building a prototype, it significantly restricts the ability to:

• recommend new tracks outside of the liked library
• analyze the full spectrum of available genres and artists
• build more advanced AI-driven discovery features
• implement deeper personalization and context-aware recommendations

For this reason, I would like to ask:

• Is it possible to obtain extended access to Spotify’s data for research or academic purposes?
• Are there programs, partnerships, or special permissions for thesis-level or experimental AI projects?
• What would be the appropriate path to request broader metadata or catalog-level access?

I believe this project aligns with the future direction of music platforms toward more intelligent, conversational, and personalized interaction with music.

Thank you very much for your time and consideration. I would be happy to provide additional details, documentation, or a demonstration of the system if needed.

Kind regards,
Hristina

[Ximzend]

Hi @HrissTsvetanova 
Thanks for sharing your project. It's clear you've put a lot of work into it already!
There are just a couple of important limitations to be aware of as you move toward the next stage:

• You mentioned future ideas like "Hey Spotify, recommend me tracks for chilling with friends at home." Spotify's Developer Policy doesn't allow building a voice‑enabled assistant that can control Spotify (section III.3).
• Several of your current and planned features rely on training AI or machine‑learning models using Spotify data or example mood detection, genre inference, and advanced recommendation logic. The policy explicitly prohibits using Spotify content or metadata to train or feed AI/ML models (section III.14).
• About your question on extended or academic access: Spotify doesn't offer special research‑level access or broader catalog data beyond what’s available through the public Web API, so there's unfortunately no path to request deeper catalog‑wide metadata.

You can definitely keep experimenting with the data the API already provides, but anything involving voice control or AI models trained on Spotify data would fall outside what’s allowed at the moment.

Wishing you the best with the rest of your thesis!