Contour

Contour Runtime 

An adaptive scoring engine that transforms game levels and physical space into a living symphony as players explore

Traditional interactive video game soundtracks are fundamentally constrained by storage media and manual sound-design tracking. In standard implementations using high-level desktop authoring tools like FMOD Studio, sound designers must map out thousands of pre-recorded audio stems. The game engine then executes basic vertical layering or horizontal sequencing to crossfade between these static audio files when a player steps across arbitrary, invisible trigger lines. This approach suffers from massive storage overhead, linear composition limits, and an immense allocation of human hours to manually plot out rule matrices.

Algorhythms Contour flips this traditional approach completely on its head by shifting game audio from reactive playback to real-time, context-aware composition. It represents the exact inverse of procedural generation: rather than using an algorithm to build a landscape, the Contour Runtime parses the live geometry and topology of the finished game world to translate the rendered environment directly into a living, nonlinear musical score.

By utilizing a deterministic, grammar-based symbolic computing pipeline instead of unpredictable, latency-heavy neural networks, Contour Studio extracts deep structural meaning from a scene’s layout. It maps physical environmental coordinates directly to compositional rules, synthesizes the output via localized SoundFont (SF2) wavetable engines, and streams raw audio data directly into the embedded FMOD Core Runtime Engine inside the game client. The result is a truly adaptive soundscape native to the world itself, achieving true zero-storage, microsecond-latency game audio completely uncoupled from fixed timelines or pre-baked audio files.

In formal grammar theory, this is known as graph parsing or an inverse graph grammar. The system scans the entire topology of the map, collapsing complex heightfields, room dimensions, portals, and navigation meshes into a clean structural tree. The physical features a player encounters: ridges, bottlenecks, wide-open plains, and dead ends are stripped of generic “mood tags” and treated as functional nodes with their own rewrite rules.

These spatial structures serve as a generative grammar for the composition. As a player moves, physical geography dictates musical form in real time:

    • Elevation pulls the melodic contour upward.

    • Branching paths thicken the underlying harmony.

    • Tight, claustrophobic geometry compresses the rhythmic pulse and narrowing range.

    • Open ground gives the melody room to breathe and lengthen.

Because a corridor and a desert do not share the same underlying graph, they can never share the same musical cue. The music doesn’t just react to the game; it is an exact sonic reflection of the space the player inhabits.


Four steps translate physical terrain into a live musical score:

    • Filter the noise. The system bypasses superficial, decorative meshes and focuses entirely on the core playable layout.

    • Extract the blueprint. It analyzes how spaces connect, mapping out bottlenecks, expansive openings, and repeating architectural motifs.

    • Bridge space and sound. That structural blueprint maps directly onto a custom musical grammar: motifs, instrumentation, chord structures, and rhythmic scales. The human composer still authors the artistic language; the level simply performs it.

    • Generate at runtime. The final graph is serialized as MIDI in real time. As the player moves, the system continuously decomposes and rebuilds the underlying structural graph, seamlessly shifting the soundtrack beneath their feet.

The resulting soundtrack is far more than background music. It is a sonification of space: a mathematical, auditory mirror of the terrain. The structural form of the world dictates the form of the music, scene by scene.

Contour fits seamlessly into existing development workflows as a plugin for Unity and Unreal. It doesn’t replace your composer or traditional audio middleware like Wwise, FMOD, or MetaSounds. Instead, it lives alongside them to handle a task they weren’t built for: parsing a level’s actual geometry to perform a human-authored, dynamically generated score.

By tapping into data the level already tracks: navigation meshes, rooms, portals, elevation, and density Contour builds its foundational blueprint. The composer defines the creative boundaries using motifs, instrumentation, and harmonic rules. At runtime, the plugin matches the player’s immediate spatial coordinates to that grammar, outputting real-time MIDI or feeding directly into MetaSounds. When the player takes a step, the underlying graph shifts, and the melody follows.

For game studios, creative control remains untouched. The studio owns the music, and the composer writes a single cohesive musical language. Contour Studio manages that generative grammar to adapt to any iteration of a level, while the engine plugin allows each scene to articulate it dynamically. This eliminates the need to record massive asset libraries for every possible coordinate a player might explore.

Beyond gaming, this framework scales to any interactive environment. Wherever a person moves through a designed physical or virtual space, Contour ensures the soundscapes reflect the architecture of the world itself, rather than looping a static, predictable playlist.


Museums, galleries, and exhibition spaces are already built on a natural spatial graph: a narrative sequence of rooms, open atria, sudden sightlines, and quiet cul-de-sacs. Similarly, theme parks rely on choreographing human traffic, moving guests through dense queues, dramatic reveals, narrow chokes, and wide plazas. Instead of forcing these dynamic transitions to rely on repetitive audio loops, this approach lets the architecture itself score the journey.

In architecture and real estate, a designer or client can walk directly through a Building Information Modeling (BIM) file or a 3D laser scan and hear the blueprint unfold as an immersive acoustic experience.
 
Extended reality (XR): spanning VR, AR, mixed reality, and location-based entertainment takes this concept even further than traditional gaming. When a user steps into a scanned warehouse or a meticulously rebuilt historic ruin, the physical environment speaks a cohesive, reactive musical grammar. The space dynamically performs its own soundtrack in real time, completely replacing the need for static, pre-compiled audio stems.
 


This same adaptive architecture applies beautifully to everyday navigation. Think of a traveler walking through an airport, a hospital, or a sprawling transit station, the soundtrack can naturally thicken into rich, layered tones at crowded junctions and thin out into quiet, minimalist textures down long corridors.
 
For urban planners and data analysts who spend all day staring at maps, digital twins and city models can suddenly be sonified. Instead of just looking at data plots, you can actually hear traffic congestion build and network connectivity shift in real time.
 
Even on virtual-production stages, where physical LED volumes and digital backgrounds change geometry between takes, the sound stays completely honest. Because the musical grammar is locked directly to the live space, the soundtrack instantly reshapes itself the moment the set moves.

Why Your Graph-Rewrite Pipeline Breaks Through the Static

Contour sidesteps traditional interactive audio limitations by translating abstract computer science concepts into three precise structural advantages that game developers are desperate to find:

  • Absolute Deterministic Control: Audio directors reject unpredictable generation. Because Contour utilizes a symbolic graph grammar instead of an erratic machine-learning model, it eliminates random variance. You can guarantee that specific scene nodes will trigger a designated musical mood, scale, or arrangement 100% of the time, providing a predictable architectural blueprint rather than a black box.
  • Frame-Accurate Telemetry: Music cannot exist in an isolated silo. By outputting a parallel, highly structured MIDI telemetry stream alongside the raw audio, Contour allows the game engine to “see” exact rhythms and notations. Technical designers can instantly bind visual shaders, lighting mechanics, or enemy pathing logic directly to the live musical frequencies of the environment.
  • Elimination of the Storage Crisis: Modern open-world games suffer from severe asset bloat, frequently exceeding 100 gigabytes. Contour solves this multi-million-dollar storage constraint by delivering an endless, non-repetitive score that adapts to player geometry while occupying under 5 megabytes of total disk footprint (compiled grammar rules plus compressed SoundFont instruments).


Contour Studio

The composer’s co-pilot for developing songs and scores

Think of Contour Studio as a co-pilot for composers. You feed it a starting point—a brief motif, a short four-bar phrase, a melodic cell, or a simple cadence and the tool goes to work, generating new melodies and harmonies from that seed on the fly.
 
By applying structural principles like graph rewriting, symmetry, and self-similarity, Contour Studio transforms your initial cell into fresh variations. It uncovers creative paths and unexpected shifts you might not have considered, all while strictly preserving your personal musical voice.
 
Ultimately, you are in complete control. You can accept, edit, or discard any suggestion. Nothing becomes part of the final piece until you say it does; Contour Studio is simply there to spark ideas, not to finish the song for you.

This is a far cry from prompt-to-song generators like Suno. Contour Studio focuses entirely on the microscopic level: the musical cell. It generates precise, high-fidelity MIDI that a composer can actually use immediately. Because you establish the underlying scale, interval rules, and instrumentation, you maintain complete creative authorship and copyright. The tool is simply a second set of ears and hands working with material you already trust.
 
You can work directly from your keyboard or feed the system a short MIDI sequence. The suggestions arrive as distinct lines and inner voices that you can drag and drop straight into your DAW. You keep what works, tweak what needs adjustment, and push the cell again the moment the creative momentum slows down or you hit a wall. The goal here is a useful, inspiring continuation, not a finished, generic cue.
 
The exact same structural logic drives both of our products. Where Contour Runtime reads data as physical land, interpreting rooms, ridges, and player paths, Contour Studio reads that same data as a musical motif, mapping peaks, valleys, and open ground. A musical period has its own bottlenecks; a melodic phrase has its own elevation. You don’t have to force yourself to pretend a song is a literal landscape—the underlying form inherently has its own unique topography.

Studio writes the language and Runtime performs it across the landscape. In the end, the topography itself sets the tone.