Building Your Music DNA: How AI Analyzes Your Listening Habits

Every person has a unique relationship with music. You might think you're "just a hip-hop fan" or "mostly into indie," but the truth is far more nuanced. When you zoom in on your actual listening patterns — the energy levels you gravitate toward, the tempos you prefer, whether you lean toward happy or melancholic music — a detailed sonic fingerprint emerges. That's your Music DNA.

I built the DNA Scan feature in Orphea because I was frustrated with how superficial most music profiles are. Your Spotify Wrapped tells you your top artist and total minutes. Cool, but what does that actually reveal? It's like describing your diet by saying "I eat food." The interesting stuff is in the details — the patterns, the preferences, the tendencies you might not even be aware of.

Music DNA is different. It's a multi-dimensional profile built from analyzing the actual audio characteristics of your library — energy, valence, danceability, tempo, and more. It doesn't care about genre labels or artist names. It cares about how the music sounds. And the results are consistently surprising, even to people who think they know their own taste inside out.

In this article, I'll walk you through exactly how the DNA Scan works, step by step. No black boxes, no hand-waving — just a clear explanation of the pipeline, the AI model, and what the results actually mean.

The DNA Scan starts by connecting to your streaming account. Orphea currently supports SoundCloud, TIDAL, and Apple Music. You authenticate through the provider's official OAuth flow — Orphea never sees or stores your password.

For SoundCloud and TIDAL, this is an OAuth 2.1 + PKCE flow, which is the most secure authentication standard available. PKCE (Proof Key for Code Exchange) means that even if someone intercepted the authentication request, they couldn't use it — the verification code is mathematically bound to your specific session.

For Apple Music, authentication happens through Apple's MusicKit JS framework, which works directly in your browser. You authorize Orphea to access your library through Apple's native authentication prompt.

Once authenticated, Orphea requests access to your library — specifically your liked/favorited tracks. It doesn't need your listening history, your playlists, or your payment information. Just the tracks you've explicitly saved or liked. This is deliberate: liked tracks represent your active choices, not passive background listening. They're a truer signal of your actual taste.

The entire authentication process takes about 10 seconds. Your provider credentials are never stored on Orphea's servers — only the OAuth access tokens, which can be revoked at any time from your provider's settings.

After authentication, Orphea pulls your liked tracks from the provider. Each provider has a slightly different API, but the data we extract is consistent:

• Track name
• Artist name
• Album name (when available)
• Track duration
• Provider-specific ID (for deduplication and caching)

For SoundCloud, this means your "liked" tracks. For TIDAL, your "favorites." For Apple Music, your library additions. Each provider uses slightly different terminology, but the concept is the same: tracks you've actively chosen to keep.

Deduplication

If you've connected multiple providers, Orphea deduplicates across platforms. The same song on SoundCloud and TIDAL gets counted once, not twice. This is done by normalizing track and artist names — stripping special characters, handling common variations (feat. vs. ft. vs. featuring), and matching on the canonical title+artist combination.

Minimum Threshold

The DNA Scan requires a minimum of 3 tracks to run. But honestly, 3 tracks gives you a pretty rough sketch. I recommend at least 10-15 tracks for meaningful results, and 30+ for a really accurate profile. The more data, the more the noise of individual tracks averages out and the clearer your true preferences become.

This is where the magic happens. For each track in your library, Orphea needs to determine its audio features: energy, valence, danceability, tempo, acousticness, and instrumentalness. How it gets those features depends on the source.

The Spotify Path (When Available)

Spotify's API provides pre-computed audio features for every track in their catalog. These are extracted from the actual audio signal using Spotify's proprietary algorithms. It's highly accurate and computationally free — just an API call. However, this path is only available for Spotify users (and Orphea's Spotify integration is currently limited to users with existing API keys).

The AI Inference Path

For SoundCloud, TIDAL, and Apple Music users, Orphea uses AI inference. Orphea's AI model receives the track title and artist name as input and predicts audio features based on its training data.

How does this work? The model has been trained on millions of songs and their associated audio characteristics. When you give it "Boards of Canada — Dayvan Cowboy," it knows from its training data that this is an ambient electronic track with moderate energy, low valence, low danceability, and a tempo around 100 BPM. The predictions are based on the model's comprehensive knowledge of recorded music.

The Analysis Pipeline

• Input: Track title + artist name
• Model: Orphea's AI (cloud AI infrastructure)
• Output: Energy (0-1), valence (0-1), danceability (0-1), tempo (BPM), acousticness (0-1), instrumentalness (0-1)
• Caching: Results stored in Orphea's database (Supabase) for instant retrieval on future scans
• Fallback: If the AI service is temporarily unavailable, previously cached results are used

Once every track has been analyzed, Orphea aggregates the results into your DNA profile. This isn't just simple averaging — the profile captures several dimensions:

Central Tendencies

The average value of each feature across your library. If your average energy is 0.72, you generally prefer fairly intense music. If your average valence is 0.38, you lean toward emotionally complex, darker-sounding tracks. These averages form the "center" of your musical taste.

Spread and Variance

How much variety exists in each dimension. Some people have a very tight energy range (always 0.6-0.8) while others span the full spectrum (0.1 to 0.95). High variance means you're eclectic in that dimension. Low variance means you have a strong, consistent preference.

Feature Correlations

How your features relate to each other. Do you tend to like high-energy AND high-valence tracks (upbeat party music), or high-energy AND low-valence tracks (intense but dark)? These correlations reveal the emotional texture of your taste in ways that individual features can't.

The Visual Profile

Your DNA is presented as a radar chart (think spider web diagram) with each axis representing a feature. At a glance, you can see the shape of your musical identity. A balanced, round shape means eclectic taste. A spiky, asymmetric shape means strong preferences in specific dimensions.

The DNA profile is stored and updates every time you run a new scan. Over time, you can track how your taste evolves — which is one of the features I'm most proud of. Unlike Spotify Wrapped (which gives you one snapshot per year), your Orphea DNA is a living, breathing reflection of your current musical self.

Since Spotify Wrapped is the most well-known music reflection tool, let me be upfront about how Orphea's DNA Scan differs and where each one wins.

Spotify Wrapped

• Frequency: Once per year (December)
• Data: Total minutes listened, top artists, top tracks, top genres
• Depth: Surface-level (counts and rankings)
• Visual: Beautiful, shareable stories
• Platform: Spotify only
• Free: Yes, for all Spotify users

Orphea DNA Scan

• Frequency: Any time, as often as you want
• Data: Audio features (energy, valence, danceability, tempo, acousticness, instrumentalness)
• Depth: Multi-dimensional analysis with correlations and variance
• Visual: Radar chart DNA profile
• Platform: SoundCloud, TIDAL, Apple Music (multi-provider)
• Free: Yes, with 5 AI credits/month on the free tier

They're fundamentally different tools. Wrapped tells you what you listened to. DNA tells you how your music sounds and what patterns exist in your preferences. Wrapped is a fun annual retrospective. DNA is an ongoing analytical tool.

One thing Wrapped does better: the social/sharing aspect. It's designed to go viral, and it does. Orphea's DNA is more personal and analytical — it's for you, not for your Instagram story (though I'm working on making it more shareable).

So you've run a DNA Scan. You're staring at a radar chart with numbers. What does it all mean? Here's a practical guide to reading your profile:

Energy Profile

• High (0.7+): You prefer music that grabs your attention. Rock, electronic, hip-hop with heavy beats. You probably use music to energize, motivate, or focus.
• Mid (0.4–0.7): You're balanced — comfortable with both intensity and calm. Most diverse listeners fall here.
• Low (below 0.4): You gravitate toward gentle, contemplative music. Acoustic, ambient, lo-fi, classical. Music is likely a calming or introspective force in your life.

Valence Profile

• High (0.6+): You prefer music that sounds happy, bright, and uplifting. Pop, funk, reggae, major-key everything.
• Mid (0.35–0.6): You like emotional complexity — music that's neither purely happy nor purely sad. This is actually the most common range.
• Low (below 0.35): You're drawn to dark, moody, melancholic music. Minor keys, tension, emotional weight. This doesn't mean you're sad — it means you find beauty in darkness.

Danceability Profile

• High (0.7+): Rhythm is central to your listening. Dance, hip-hop, pop, house. You probably listen to music in social or physical contexts (parties, workouts, driving).
• Low (below 0.4): You listen to music for texture, melody, or lyrics more than rhythm. Progressive rock, ambient, classical, singer-songwriter.

Tempo Profile

• Below 90 BPM: Hip-hop, R&B, downtempo, chill. Relaxed pace.
• 90–120 BPM: Pop, rock, indie. The "comfortable" range for most Western music.
• Above 130 BPM: Electronic, punk, drum and bass. You like things fast.

Your music taste isn't static. It evolves with your mood, your life circumstances, the seasons, the people around you. That's why I designed the DNA Scan as a repeatable tool, not a one-time test. Run it every few months and you'll see how your sonic fingerprint shifts and develops over time.

Some things will stay remarkably consistent — most people have a core energy range and tempo preference that barely budges. But your valence might swing with life events. Your danceability might increase during summer and decrease during winter. These shifts are natural and fascinating to track.

The DNA Scan is also the foundation for everything else in Orphea. Your Music DNA powers the recommendation engine, influences what The Cut shows you, and helps you understand why certain tracks resonate and others don't. It's not just a profile — it's a lens for understanding your relationship with music at a deeper level.

Music is one of the most personal things in our lives. Understanding why you love what you love doesn't diminish the magic — it deepens it. Your DNA is the map. Now go explore it.