Vertical layering keeps one piece of music playing and fades stems in or out as game intensity changes, so the music thickens without ever restarting. Horizontal resequencing switches between separate sections or loops at musical boundaries, so the music changes character. Use layering for gradual tension and resequencing for clear state changes; most good games combine both.
Last reviewed: October 2026
What is vertical layering?
Vertical layering (also called vertical remixing or stem mixing) means every layer of the arrangement plays in sync from the start, and the game controls which layers are audible. A typical exploration cue might have pads, a bass line, light percussion, a melody and a high-energy rhythm layer. At low intensity only the pads and bass are audible. As danger rises, the percussion fades in, then the melody, then the driving rhythm.
Because all the stems are the same length and share one clock, they can never drift apart. The player experiences one continuous piece of music that gets denser or sparser, which makes it ideal for things that change smoothly: enemy proximity, health, the number of enemies on screen, or distance to an objective.
The main design constraint is that every layer must work with every combination of the others. You are composing a chord progression and tempo that must sound complete with only the bass and sound complete with everything. That tends to push you toward simple, strong harmonic foundations rather than complicated modulations.
What is horizontal resequencing?
Horizontal resequencing (also called horizontal re-sequencing or segment-based music) means the music is built from separate sections, such as an explore loop, a combat loop and a victory section, and the game chooses which one plays next. The switch happens at a musically sensible moment, usually the next beat, the next bar or the end of a phrase, so the change feels intentional rather than like a cut.
This is the right tool when the emotional character changes, not just its intensity. Moving from a quiet village to a battle usually means a different tempo, a different harmonic area and a different rhythmic feel. You cannot get there by fading in a stem; you need a new section that was written to be that thing.
The cost is transitions. Every pair of sections that can follow each other needs either a clean boundary or a bridging piece, and the number of pairs grows quickly as you add states. If you want the foundations, our guide to states and transitions covers how to keep that manageable.
Vertical layering vs horizontal resequencing: how do they compare?
| Aspect | Vertical layering | Horizontal resequencing |
|---|---|---|
| What changes | Which stems are audible | Which section is playing |
| Best for | Gradual intensity (tension, proximity) | Distinct states (explore, combat, victory) |
| Tempo and key | Fixed for the whole piece | Can differ per section |
| Transition risk | Very low, stems stay in sync | Needs good boundaries or bridges |
| Composition cost | Every layer must fit every mix | Each section needs to be a finished loop |
| Memory and CPU | All stems play at once, so more voices | Usually one section plays at a time |
| Typical trigger | A continuous value, like danger from 0 to 1 | A discrete event, like combat starting |
| Main weakness | Limited emotional range | Audible seams if transitions are poor |
The memory and CPU row matters on mobile and in web builds. Streaming eight stems simultaneously costs more than streaming one section, so check your platform budget before you commit to a large layer count.
When should I use each technique?
Ask what the player's situation is doing. If it moves along a smooth scale, use layers. If it jumps between categories, use sections.
- Stealth, survival and exploration games with a continuous threat level lean on vertical layering.
- Action games with clear combat encounters lean on horizontal resequencing: an explore loop, a combat loop, a short stinger when the fight ends.
- Boss fights are a classic hybrid, covered below.
- Puzzle and cozy games often need neither beyond a few layers that reward progress. See our notes on puzzle game music.
If you are unsure, start with layering. It is the cheaper mistake: the worst case is that your music is a little less dramatic, whereas badly handled resequencing produces audible seams.
How do I combine both in a hybrid approach?
Most shipped adaptive scores are hybrids. The common pattern is to use horizontal resequencing at the state level and vertical layering inside each state.
For example, a game might have three states: Explore, Combat and Victory. Inside Explore, a handful of layers respond to how close the player is to danger. Inside Combat, extra layers respond to how many enemies remain. The transition between Explore and Combat is a resequencing step, scheduled at the next bar so it lands on the beat.
Two practical rules keep hybrids sane:
- Share tempo and key between states that transition often. If Explore and Combat are both 100 BPM in the same key, the switch can happen on any bar and still sound composed. If they differ, schedule switches on phrase boundaries and consider a short stinger to hide the join.
- Keep layer counts modest. Four or five layers per state is usually enough for a player to feel the music responding. Beyond that, individual layers become hard to hear as distinct changes.
This is also how Mowjera structures a score. You define states, each a loop of whole bars, and transitions between them can be immediate, next bar or next phrase. Within a state, tracks are layered by an intensity threshold (a minIntensity per track), so a track only becomes audible once the intensity value reaches its threshold.
How do transitions work in each technique?
In layering, the "transition" is a volume ramp. Fading a stem in over one or two bars usually sounds natural; fading it in over half a second can sound like a mixer move rather than music. Fade on bar lines when you can, and fade out more slowly than you fade in so the arrangement does not feel like it is flickering.
In resequencing, the transition is a scheduling decision. The three common options are:
- Immediate. Switch right now. Use for urgent events like a sudden ambush, accepting that it may not land on a beat.
- Next bar. Wait until the current bar ends. A good default for most state changes.
- Next phrase. Wait for the end of a longer unit, typically four or eight bars. Best when the music has a clear phrase shape you do not want to interrupt.
Be careful with reverb and delay tails. If the outgoing section has a long tail, either let it ring over the new section or cut it deliberately; the worst outcome is an accidental half-cut. Our piece on making seamless game music loops goes into tail handling.
Which tools support these techniques?
Both techniques are standard in middleware. Wwise and FMOD are excellent tools with deep support for layered and sequenced music, and if you have a sound designer on the team and a large project they are a strong choice. See Wwise alternatives and FMOD vs Mowjera for a fair comparison. You can also build either technique directly in an engine; our Godot tutorial and Unity guide show how.
If you want to author the music itself without setting up middleware, Mowjera is a browser-based DAW built around this model. It exports per-state looping WAV files, a metadata.json, and generated controller scripts for Godot 4, Unity and Unreal. You can try it without an account.
FAQ
Is vertical layering better than horizontal resequencing?
Neither is better in general. Vertical layering suits smooth changes in intensity and is low-risk because stems stay in sync. Horizontal resequencing suits distinct states with different tempo or key. Most games use resequencing between states and layering within each state, because the combination covers both gradual and abrupt changes.
What is the difference between vertical and horizontal adaptive music?
In vertical adaptive music, multiple synchronized layers play together and the game raises or lowers their volume to change intensity. In horizontal adaptive music, the game switches between separate sections of music at musical boundaries. Vertical changes density; horizontal changes character, tempo or key.
Do I need Wwise or FMOD to use these techniques?
No. Both techniques can be implemented with a few lines of code in any engine that can play synchronized audio, and middleware simply makes it more convenient at scale. Smaller projects and game jams often do well with a simple state machine and pre-exported loops.
How many layers should a vertically layered track have?
Four to six layers is a common range. Fewer than three gives little sense of change, and many more tend to blur together so players cannot hear individual changes. Design each layer to be audibly distinct, such as pads, bass, percussion, melody and a high-energy rhythm layer.
Should different states share the same tempo?
If they transition often, yes, it makes switching on any bar sound composed. States that differ in tempo are fine but need transitions on phrase boundaries or a stinger to cover the join. It is a design trade-off between musical contrast and seamlessness.
Try it on your own score
If you want to hear the difference rather than read about it, build a short Explore state with three layers and a Combat state, then switch between them at the next bar. You can start free in Mowjera, or read the beginner's guide to adaptive music first.