Death Coming Up The Hill

Death Coming Up The Hill: How to Actually Make It Work

Most people try to build this sound by stacking sawtooth waves and sliding the filter up. That gets you close, but it sounds generic. The real issue is that the technique relies on harmonic behavior most beginners ignore until their mix falls apart. I'll walk through the actual signal path, then explain the definitions as we go.

Building the Core Signal

The foundation is not a single oscillator. It is three layered sawtooth patches, each running through its own resonant lowpass filter. The trick is detuning them slightly against each other — about 8 to 12 cents — so the harmonics don't phase-cancel into a thin line. If they do, you hear it immediately when you automate the filter cutoff upward. The filter automation is where the sound lives. You open the cutoff from around 200 Hz to somewhere near 8,000 to 12,000 Hz over roughly 2 to 4 seconds. The resonance needs to be set between 0.7 and 0.85. Anything higher and you get a nasty squeal right before the peak hits. Anything lower and the harmonic overtone stack never really develops, which is the whole point of the technique. On my end, I use Serum for the basic patch. It has a built-in riser preset that does most of the heavy lifting, but I always strip it down and rebuild it from scratch. The preset layering is too clean, too symmetrical. Real Death Coming Up The Hill has an irregular harmonic curve where some partials hit harder than others. I achieve this by adding a sub-oscillator an octave down at 30% volume, then running everything through a bitcrusher set to about 12-bit at 24 kHz sample rate for 200 milliseconds of duration. That adds the grit without muddying the low end.

Here is where it gets complicated, and where most tutorials fail you.

The Harmonic Distribution Problem

The overtone series in a sawtooth wave follows a 1/n amplitude distribution. That means the 2nd harmonic is half the amplitude of the fundamental, the 3rd is a third, and so on. When you sweep a lowpass filter upward, the harmonics don't all become audible at the same rate. The early partials (2nd through 5th) come through first and can dominate the midrange. The higher partials only reveal themselves near the end of the sweep, and they tend to sound harsh because they are hitting the Nyquist limit on your DAC. The workaround I ended up using after wasting several days on this was a multiband compression stage right after the filter. I split the signal into three bands: 200 to 2,000 Hz, 2,000 to 6,000 Hz, and 6,000 to 12,000 Hz. Each band gets its own compressor with a fast attack (5 ms) and a release time that follows the automation envelope. This prevents the midrange from peaking during the middle of the sweep and lets the high harmonics bloom at the end without clipping. Without this stage, the sound either collapses in the middle or clips at the top. Both outcomes are unusable in a final mix.

Common Pitfalls and What Actually Happens in Practice

The biggest mistake I see is ignoring the stereo field. People process this sound in mono and then wonder why it feels small when they drop it into a stereo mix. The fix is simple: send the left and right channels through slightly different filter Q values. Left channel at 0.75 Q, right at 0.82. This creates a subtle width that makes the riser feel like it is approaching from a specific direction rather than sitting dead center. Another thing that catches people out is automation latency. If you are using a DAW with significant plugin delay compensation enabled, your filter sweep might not line up with your visual timeline markers. I learned this the hard way on a project where the audio was always slightly behind the animation frame. The fix was disabling delay compensation on the individual track and routing it through a manual delay-compensated bus instead.

When This Technique Fails Completely

Death Coming Up The Hill does not work well for short impacts. If your cue is less than 1.5 seconds, the harmonic buildup never has time to develop, and the sound just becomes a noisy spike. For those cases, use a noise burst with a high-pass filter sweep instead. It is faster to produce and sounds more natural for quick stingers. It also struggles in dense mixes. If you have bass guitars, kick drums, and multiple synth pads already occupying the 200 Hz to 4,000 Hz range, this riser will get buried regardless of how much EQ you carve out. The fundamental problem is that the technique occupies the same frequency space as most other midrange elements. Sidechain compression from the kick or bass onto the riser bus can help, but it makes the sound pump unintentionally. The cleaner solution is to high-pass the riser at 150 Hz and accept that you are losing some body in favor of clarity.

A Quick Reference Path

Here is the exact chain I use when I need this sound fast: Oscillator 1: Sawtooth at 1x fundamental Oscillator 2: Sawtooth at 1x fundamental, detuned +10 cents

Oscillator 3: Sawtooth at 1x fundamental, detuned -10 cents Sub oscillator: Square wave at 0.5x, volume at 30% Filter: Lowpass, cutoff automation 200 Hz to 10,000 Hz over 3 seconds, resonance at 0.78

Multiband compressor: 3 bands, attack 5 ms, release synced to automation Bitcrusher: 12-bit, 24 kHz, 200 ms duration Reverb: Plate, 1.2 second decay, 25% wet, send only

This setup takes about 15 minutes to program once you know the parameters. The first time through, expect closer to 45 minutes because you will be tweaking the Q values and checking for phase cancellation in mono. The sound itself is essentially a controlled harmonic ascent. The name comes from the feeling it creates — something getting closer, inevitable, building pressure until it either resolves or cuts off abruptly. That is all there is to it. It is not magic, it is just understanding how harmonics behave under a moving filter and having the discipline to compress the result so it does not blow up your mix.