HIGHER Audio Direction

HIGHER should sound like a playful collection of physical toys, unreliable machines, and surprisingly dangerous birds. The recommended palette is a compact pop for gunfire, a dry mechanical clack for weapon identity, clear tonal signals for rewards and warnings, and composed funk/fusion music that gives fights a mischievous rhythm. Loudness is a supporting tool; impact comes primarily from timing, body, contrast, and the relationship between sound and animation.

The implemented revision replaces both gun families, adds three variants per gun, and introduces different licensed music for the two maps. It retains the existing short crown cues, chicken calls, ignition loops, impact sounds, doors, and breakable-platform cues. Independent effects and music faders, bounded voices, pause handling, result ducking, and musical continuity across rematches are part of the same rollout.

Duck Game reference

Landon Podbielski described Duck Game as an accessible arena game with short rounds and immediately usable weapons. In a later interview, he connected pitched quacks, hats, ragdolls, and post-round gloating to player expression and slapstick. The transferable lesson is that sound should support amusing physical interactions and recognizable personalities. ¹²

The official Steam page places conventional guns alongside saxophones, nets, magnetic weapons, and other improbable equipment. Adult Swim’s original trailers are useful primary reference material for the public-facing tone. They are references, not sources of reusable audio. ³⁴⁵

No reproducible Duck Game gun-recording or mixing recipe was established from those primary references. They do not document microphone choice, layering, equalizer settings, loudness targets, sample-variation rules, or the game’s underlying audio middleware. Claims that its gunshots use a particular commercial library, human mouth sounds, or a specific synthesis tool would therefore be speculation.

For HIGHER, the relevant design interpretation is physical comedy with useful feedback. A gun should sound sturdy enough to satisfy, but small and peculiar enough to belong in a world where touching a chicken can launch a bird across a room. Copying a military gun recording at full brightness would undermine that tone. Copying Duck Game’s recordings would also fail to establish HIGHER’s own identity.

This distinction matters because HIGHER now has health as well as knockback. A shot is often one step in an exchange rather than an immediate elimination. Ordinary hits must remain comfortable when repeated, while a shield break, crown loss, chicken explosion, or final knockout deserves extra space. The sound hierarchy therefore needs to communicate both continuous action and exceptional moments.

Current sound diagnosis

Gunfire needs an identifiable attack and a brief physical body underneath. Too much low-frequency synthesis obscures the weapon’s texture and makes a series of shots feel muffled. The final design therefore gives the recorded pop and dry mechanical detail priority, while keeping a much smaller body underneath.

The current replacement addresses the source character first. The blaster now contains a brief, filtered recording of a 1911 shot; the automatic uses a brief Carl Gustav M45 recording. Those recordings are source material for a stylized composite. The recording is the leading layer, supported by a dry wood impact, a short mechanical clack, a small muted punch layer, and a restrained synthesized body. This preserves the irregular texture of a physical action instead of letting a drum-like low tone define the whole weapon.

The intended result is a compact, toy-like pop with a distinct mechanical edge: a larger pop-and-ka-chak for the blaster and a smaller, quicker rattle for the automatic. It avoids the long, airy tail of a distant explosion and the high, melodic contour of a laser. The game’s actual lasers retain their own electronic identity, which helps distinguish a weapon exchange from a lethal environmental threat.

At the mix level, a six-player firefight is different from a single shot auditioned in isolation. Six equally prominent automatic weapons can create a nearly continuous band of noise. The revised mixer continues to prioritize the local player, attenuates ordinary bot effects, and caps simultaneous gun voices. The purpose is to preserve responsiveness without requiring every overlapping remote shot to remain a full-volume foreground event.

Gun construction and measurements

Each gun family has three prepared samples. Variants change the blend of the small punch and synthesized body and adjust the recording’s low-pass cutoff between 6.2 and 6.7 kHz. A dry wood layer enters 2 ms after the primary onset. The metal clack follows at 22–25 ms for the automatic and 55–58 ms for the blaster. In play, small pitch and gain offsets add further variation without changing fire rate or consuming the game’s random-number sequence.

The Free Firearm Sound Library identifies its authors and explicitly permits personal and professional use under CC0. Its archive includes a master sheet describing the firearm and recording position. Only a fraction of a second from two recordings is present in the shipped composites; the 194 MB source archive is excluded from the game. ⁶

The Kenney punch and metal-click recordings were already part of the project’s licensed source set. Kenney marks Impact Sounds as CC0. The small body layer is an original decaying sine construction, shifted above the deepest bass and shortened to leave the attack clear. Its blend weight is approximately 73% lower than the earlier thud-led mix; the muted punch weight is approximately 77% lower. It is not sampled from a song or another game. ⁷

These figures describe the delivered files and configured mix. They are not subjective listening scores. Frequency-energy share confirms that sub-bass is restrained and the attack has a separate upper-mid identity, but it does not by itself prove that a listener will find it satisfying. The blaster’s RMS level is unchanged and the automatic is slightly quieter; the revision gains clarity through balance rather than volume. The important comparison is a complete magazine at the normal game volume, with movement, music, and other players active.

The automatic’s 145 ms sample is short enough to overlap mainly the next shot at the existing roughly 110 ms interval. The source is started once for each accepted shot; there is no long looping gun recording that can drift away from the visible muzzle flashes. The mixer uses a short additional release and stops obsolete tails on pause or reset.

Pitch variation remains deliberately small: approximately −1.4% to +1.8%, with a five-step sequence. This is enough to reduce a repeated-sample quality without turning successive shots into a melody. The three source variants cycle per weapon and shooter, so one bot’s firing cannot change the local player’s next sample choice.

Candidate comparison

The preferred source is not necessarily the most realistic source. For this game, recording provenance, a clean editable onset, short-tail suitability, and room for stylization matter more than firearm authenticity. A field recording with a documented author can be useful even when the final result bears little resemblance to the source weapon.

The selected approach avoids an unnecessary accumulation of libraries. A consistent family built from a few well-documented ingredients is easier to balance than dozens of unrelated recordings. The same principle applies to future additions: a weapon should earn its own sonic identity through a clear function, not merely through a louder or longer sample.

Music selection

Two existing compositions were selected over a new procedural score. They provide longer musical ideas, changing accompaniment, phrasing, and stylistic personality that a short generated ostinato would struggle to match. Both are published as CC0 by their creators on OpenGameArt, with directly offered source files. ¹¹¹²

Tower — Raspberry Jam, by congusbongus. The creator identifies it as a New Jack Swing tune made in MilkyTracker and supplies the tracker source. Its rhythmic identity is a good analytical match for an open arena with quick movement and recoveries. HIGHER uses a 137.143-second game edit, preserving a 64-bar span at the supplied source’s 112 BPM. ¹¹

Dead Letter Depot — Rhythm Factory, by Zane Little Music. The creator describes an intentionally unusual composition with changing ideas and sounds. That makes it a stronger candidate for a mechanical postal station with doors, parcel cover, and a failing express bridge. HIGHER uses a 118.154-second editorial game adaptation of the public composition. The creator’s paid Deluxe loops and stems are not used. ¹²

The two tracks are related enough to belong to the same playful game but different enough to give the maps distinct identities. Neither needs to react to every shot. Their role is to sustain motion beneath the fight and make quiet intervals feel inviting, while the event mix retains responsibility for danger, contact, and reward.

The selections are creative recommendations supported by source descriptions, licensing, and technical inspection. They are not a claim that the composers designed these tracks for HIGHER or endorse the game. The final sound desk makes both delivered edits available for listening in the actual mixer.

Music edits and playback

Both scores are stereo 44.1 kHz MP3 files encoded at 160 kbps. Their combined transfer size is approximately 5.11 MB. They load separately from the much smaller effects bank, and only when play actually requests a map score. A slow music request therefore cannot hold up gunshot or movement-effect readiness.

The edits retain long sections of each composition. A 350 ms post-roll wrap is blended over the beginning of the chosen region while preserving the region’s duration. This detail is important: simply overlapping and removing the end of a musical phrase would shorten its timing on every repetition. The decoded browser buffer also receives a tiny 3 ms edge envelope because a lossy encoder can change exact boundary samples.

The loudness measurements above are from the delivered encoded files before in-game volume. They distinguish perceived programme level from sample peak rather than treating the loudest individual sample as the whole story. EBU R128 supplies the relevant loudness and true-peak terminology; its broadcast target is not adopted as a mandatory game-mix target. ¹⁷

Web Audio buffer sources provide explicit looping points and sample-scheduled playback. Gain automation supplies the fades, and the music source is independent of frame-by-frame game simulation. This fits a browser brawler: camera pauses, low frame rates, or a slow-motion visual finish should not accidentally change the pitch of the soundtrack. ¹⁸¹⁹

The music engine remembers the current position when paused or muted. A same-map rematch keeps the composition moving instead of replaying its opening repeatedly. A change of map fades the outgoing source into the new one over 800 ms, and rapid changes remain bounded to one incoming and one departing music voice.

Mix hierarchy and comfort

The in-game music slider defaults to 25%, followed by a 0.55 calibration on its own bus. Effects retain their independent fader. This leaves music present during movement and preparation without asking every weapon or warning to overpower a full-strength soundtrack. Both paths share a final limiter, but changing the effects slider does not change music volume.

Important warnings, crown moments, explosions, ash events, and result cues can lower music by a further factor of 0.55, approximately 5.2 dB. Ordinary gunfire does not repeatedly pump the score. The dip holds briefly after an important cue, with a quicker reduction and a slower recovery. Results lower the music to 42% of normal under the win/loss sting.

This is an event-based ducking arrangement. Audiokinetic’s documentation describes the same general concept of temporarily reducing one bus to bring another signal forward, with separately controlled fade and recovery timing. HIGHER uses its own small Web Audio implementation; no claim is made that it includes Wwise or reproduces Duck Game’s mixer. ²⁰

Routine effects remain bounded to twelve active voices, with category limits for gunfire, impacts, movement, pickups, and other groups. Two flame loops are allowed, prioritizing the player and the closest relevant opponent. Chicken idle calls remain sparse and yield to more important events. Quiet feet and prop details should make the world feel tactile without turning a quiet second into another mandatory announcement.

The strongest control for comfort is still the ability to choose a suitable listening level. No software gain, peak measurement, or limiter setting can guarantee a comfortable physical loudness across all devices and headphones. The game therefore exposes separate effects and music controls and keeps all-audio mute immediate, while fades prevent the control itself from adding a click.

Event identity and comic timing

Gunfire is only one part of the palette. The chicken’s warning, hatch, sparse idle clucks, and blast form a small comic sequence: anticipation, reveal, tension, and consequence. Its sounds should leave enough silence between calls for the player to notice the next movement or danger cue. Increasing the number of chickens should increase tactical chaos more than it increases continuous vocal noise.

Crown gain and crown loss have different directions and durations. The short loss cue ends before a long chain of unrelated impacts can make it seem like an alarm. Ash retains a brief hiss followed by a delayed rustle, matching the change from a living character to falling remains. Those sequences are more distinctive than assigning the same generic impact to every defeat.

Doors and the express bridge use physical, localized sounds. Opening and closing have separate cues; different doors can sound independently, but repeated updates of one transition do not create duplicate noises. Bridge strain comes before collapse. In a room-based arena, these cues help the map feel responsive even when the player is looking elsewhere.

The next useful sound additions should follow visible, meaningful animation. A hat settling after a knockout, a short surprised chirp at a near miss, or a small triumphant wing flourish could reinforce personality. They should be sparse, linked to an actual state change, and subordinate to hazards. A continuously talking announcer or a quip on every hit would make the game harder to read and faster to tire of.

Verification and evaluation

The automated checks cover decoded effect formats, exact effect edge samples, source hashes, metadata, payload limits, gun-family variation, short automatic tails, event duplication, category priorities, and the music lifecycle. A full automatic magazine is checked for one sound per actual shot, while simultaneous bot voices must not suppress a new local shot. These checks protect responsiveness and resource bounds rather than attempting to assign a subjective fun score.

Music checks cover gesture-adjacent loading, same-map continuity, pause/resume position, map crossfades, rapid map switching, result ducking, mute, suspended contexts, late downloads, missing files, and license/hash metadata. Late ragdoll contacts remain audible after a scored result, and the final ash rustle follows the extra 258 ms introduced by the short slow-motion finish. A failed or late music request must leave effects and gameplay usable. A completed download must never start a score behind a pause screen on its own.

The sound desk at test/audio-polish-preview.html uses the production mixer. It includes a blaster shot, automatic magazine, six-player volley, laser warning, chicken blast, crown gain/loss, and ash collapse. It also exposes map selection, independent volume controls, pause, a result transition, and same-map rematch. The page has no rank or profile persistence.

Listening evaluation should compare complete interactions. First compare the blaster and a full automatic magazine at the default volume. Then add the map music, a chicken explosion, and a warning during a six-player volley. A successful mix leaves the local shot recognizable, the hazard understandable, and the music perceptible again after the busy moment.

Repeated listening also matters. Ten seconds can favor an exaggerated sound that becomes irritating over ten minutes. The useful acceptance questions are whether rapid fire remains clear and comfortable, whether the music can continue across multiple rounds without the opening becoming repetitive, and whether the crown and chicken cues remain funny without dominating the entire session. Those are listening judgments still worth collecting from real play, even after all technical checks pass.

Rights and reproducibility

CC0 allows copying, adaptation, redistribution, and commercial use without a permission request, subject to rights the affirmer does not control. The published dedication does not imply creator endorsement. Each acquired source is therefore recorded by creator, exact page, offered download, file size, and SHA-256. ²¹

The effects manifest records source excerpt windows, processing, durations, measured levels, and final hashes. The music manifest records the source intervals, edit description, wrap treatment, encoding, measured levels, and final hashes. The credits ledger distinguishes the old base preparation from the current polish pass so rebuilding the original assets does not silently restore the previous laser-like blaster.

The current preparation is reproduced by running scripts/prepare-audio.py for the original source set, then scripts/prepare-polish-audio.py with both the original source directory and the new research-download directory. The latter contains the publicly offered music files and extracted firearm archive. Source archives remain outside the distribution; only the short composites, edited scores, metadata, and credits ship.

This revision establishes a consistent direction rather than attempting to impersonate another game. The core is a compact recorded pop, restrained body, expressive physical details, composed music with personality, and enough restraint for surprising moments to stand out. Future audio should be judged against that direction and against complete play sessions, not against isolated loudness.

Sources

All linked pages were checked on September 11, 2026. Source dates below are publication dates when available; undated technical pages are identified as such.

1. John Vanderhoef / GameCritics. An Interview with Duck Game Developer Landon Podbielski. October 16, 2014. Direct developer interview; accessibility and short-round intent.
2. Tom Brewster / The Edge. A Conversation with Duck Game’s creator, Landon Podbielski. March 12, 2019. Direct developer account of expression and slapstick.
3. Landon Podbielski / CORPTRON. Duck Game on Steam. Live product page. Public positioning and weapon variety.
4. Adult Swim. Duck Game Trailer. March 19, 2015. Official reference video.
5. Adult Swim. DUCK GAME Launch Trailer. June 4, 2015. Official reference video.
6. Ben Jaszczak, Brian Nelson, Kevin Heras, Matthew Nanney; archived by bart. The Free Firearm Sound Library. March 9, 2014. Recording provenance, archive and CC0 statement.
7. Kenney. Impact Sounds. 2019 release. Original asset page and CC0 designation.
8. Tabasco. Gunshot Sounds. July 2, 2010. Recording description and stated limitations.
9. kurt. Gunshots. February 10, 2017. Alternative recordings and echo caveat.
10. dklon. Gunshots!. December 13, 2020. Synthetic source and attribution license.
11. congusbongus. Raspberry Jam. September 27, 2020. Composer description, CC0, OGG and tracker source.
12. Zane Little Music. Rhythm Factory. July 8, 2023. Composer description, public CC0 composition, separate Deluxe offering.
13. Zane Little Music. Wednesday Night — Funk Fusion. August 28, 2023. Alternative composition and CC0 designation.
14. Joth. Funked Up. October 25, 2019. Looping track and creator’s suggested use.
15. Kevin MacLeod. Sneaky Snitch. Uploaded November 25, 2010. Instrumentation, tempo and CC BY 4.0 terms.
16. Deva. Xenocity — Sidekick. August 23, 2015. CC0 label and commercial-contact text.
17. European Broadcasting Union. R128: Loudness normalisation and permitted maximum level of audio signals. Current publication page. Loudness and peak terminology.
18. W3C. Web Audio API. Recommendation, June 17, 2021. Buffer-source and gain-automation behavior.
19. MDN contributors. AudioBufferSourceNode: loop property. Updated April 10, 2025. Loop markers and playback behavior.
20. Audiokinetic. Defining the properties of a bus — Ducking signals. Undated documentation. Ducking, fades and recovery concepts.
21. Creative Commons. CC0 1.0 Universal deed. Current deed. Public-domain dedication, permitted uses and limitations.