Most beginner indie games do not have a sound design problem. They have a small set of specific, repeatable errors that produce a recognizable result: the game feels stiff, the audio is fatiguing after ten minutes, and one sound inexplicably drowns out everything else. The errors below are the ones that show up most often, and each one has a symptom you can hear in the first minute of play.
They are ordered roughly by how often they appear, not by severity. Some are cheap to fix and some require a code change, but all of them are worth fixing before the game ships, because the player experiences them as “the audio is bad,” not as seven separate problems.
Mistake 1: One Sound File Played on Repeat Until It Becomes a Machine Gun
This is the most common error, and it has a name. In psychoacoustics, the machine gun effect describes what happens when the same short sample is triggered repeatedly: the ear stops hearing separate events and starts hearing a single, unnatural stream . A sword swing that sounds fine once becomes a mechanical rattle after the tenth swing, because the identical waveform is arriving at the same pitch and the same level every time.
Human auditory memory for short sounds is short. The working estimate from game audio practice is that the ear remembers a sound for roughly seven seconds, which means any effect that can repeat within that window needs more than one variant . For a basic sword attack, a common recommendation is at least four to five different samples, played at random with no immediate repeat of the previous one .
If there is no budget for multiple recordings, the fallback is randomized pitch and volume within a narrow range. The sound variation system guide covers the architecture; the beginner-level fix is simpler: add a small random pitch offset and a small random gain offset on every trigger, and the machine gun effect dissolves without new assets. The controlled randomness guide covers how far the offsets can go before the sound stops reading as itself.
Mistake 2: No Limit on How Many Copies of the Same Sound Can Play
The symptom is specific: a scene with many identical emitters — a chicken farm, a swarm of enemies, a wall of turrets — produces a sound that is not louder but thicker, as if the audio has turned into a single undifferentiated roar. The cause is that every emitter triggered its own copy, and the copies summed.
The same source that documents the machine gun effect gives a concrete example: a farm game where each hen clucks when laying an egg. With one hen, the cluck is charming. With 32 hens laying simultaneously, the speakers “explode” — not because any single cluck is too loud, but because 32 identical clucks are summing into a single distorted event . Even if the hens are staggered by a few seconds, the overlapping clucks become “a cacophony” rather than a farmyard.
The fix is a voice limit on the sound, not a lower volume. A counter increments on each trigger and decrements on each finish; when the limit is reached, the oldest (or farthest) instance is stopped . In Unity, this is often implemented with a small manager that tracks active instances and destroys the oldest when the cap is exceeded. The sound prioritization and voice limiting guide covers the per-object and global layers; the beginner version is that a single cap on the loudest repeating sound is better than none.
Mistake 3: Reverb That Does Not Match the Space
The symptom is harder to name than the first two, but it is audible as a mismatch: a gunshot that sounds like it was fired in a cathedral inside a small concrete room, or a footstep in a forest that carries an indoor slap-back. The cause is that the reverb was baked into the source file, or applied globally, without regard to the space the sound is playing in.
The working rule is that reverb should be a property of the space, not of the file. If the game has indoor and outdoor areas, the same dry footstep should receive different reverb depending on where the player is, which means the source should be exported without reverb and the engine should apply it through a send or a spatial audio system. The ambient background sound effects guide covers how a scene’s space is established through layers; the beginner error is treating reverb as a fixed character of the sound rather than a property of the environment.
There is a second version of this error: reverb that is too long for the sound’s role. A UI click with a 2-second reverb tail is not “atmospheric,” it is a click that will still be decaying when the next click arrives. The SFX duration guide covers the category ranges; the reverb rule is that the tail should not outlast the category’s expected repeat interval.
Mistake 4: Everything at the Same Volume, Including the Music
The most common beginner mistake, according to a game audio guide written specifically for indie developers, is making everything the same volume. Music and sound effects compete for the player’s attention, and if they are at the same level, the sound effect that is supposed to confirm a hit is buried under the background track .
The hierarchy that works is not a matter of taste. Music should sit behind SFX, because the SFX is carrying feedback and the music is carrying mood. A hit sound that is quieter than the music is a hit sound the player will miss during a busy section. The audio mix hierarchy guide covers the full bus structure; the beginner version is a simple two-level rule: SFX above music, and UI above ambience.
A related symptom is what happens when every sample is normalized to the same peak. A quiet UI click and a loud explosion can both peak at 0 dBFS, which means the click is now as loud as the explosion in peak terms, and the explosion has lost its headroom . The equal loudness guide covers why peak normalization is the wrong target; the beginner fix is to normalize by perceived loudness, not by peak, or to leave headroom and set levels by ear in the mix.
Mistake 5: Clipping That Nobody Noticed Because It Only Happens When Sounds Combine
Individual samples can all peak below 0 dBFS and still produce clipping when they play together. A USound synthesis handout lists this as a “commonly seen issue” in games: sounds are often normalized to full scale, and multiple samples at 0 dBFS usually means clipping has occurred . The clipping is not always obvious on a single sound; it becomes audible when a gunshot, a footstep, and a UI click land on the same frame.
The symptom is a “metallic” or “fizzy” quality on loud moments, or a sound that seems to flatten rather than get louder. A Godot forum thread describes the same problem from the other direction: if the sounds are distorting, the combined volume has reached the maximum, and the fix is to set levels properly and compress the bus if necessary, rather than to push the master fader . The over-processing guide covers the phase and gain-staging checks; the beginner version is to leave 3–6 dB of headroom on the master bus and to check the peak level with several sounds playing at once, not one at a time.
Mistake 6: Sound That Arrives Late or Ends Early Relative to the Animation
Timing errors are harder to hear as “audio mistakes” because they read as “the game feels unresponsive.” A jump sound that plays one frame after the character leaves the ground, a hit sound that starts before the impact frame, or a shoot sound that is shorter than the muzzle flash — each one makes the game feel disconnected even when the physics are correct.
The frame-rate dependency is part of the problem. At 30 fps, one frame is roughly 33 ms, and event timings get rounded to frame boundaries, so a sound that is “close” can land 33 ms early or late . At 60 fps the granularity is finer, but the rounding still happens. The practical fix is to place the sound on an animation event or a notifier at the exact frame of contact, rather than relying on a delay or a cooldown that may not align with the animation.
The reverse error is a sound that is the wrong length for the animation. A shoot sound that is 200 ms long paired with a 100 ms muzzle flash will continue after the visual has finished, which makes the weapon feel lighter than it looks. The jump and landing guide covers the attack-and-landing timing; the beginner rule is that the sound’s attack should coincide with the visual contact frame, and the sound’s duration should be within roughly one category step of the animation’s length.
Mistake 7: Silence Where the Player Expects Feedback
The last error is the absence of a sound, and it is the one most likely to be defended as a design choice. A menu that opens silently, a button that clicks without a confirm sound, a pickup that disappears without a chime — these do not read as “minimalist.” They read as “unfinished,” because the player has no confirmation that the input registered.
The principle from game feel research is direct: every player action should have an audio response, and if there is only time for one type of audio, SFX should be the priority . A UI click is the cheapest possible confirmation and the one the player hears most often. The game UI sound design guide covers the hierarchy between confirm, cancel, and error; the beginner version is that a confirm without a sound is a confirm the player will press twice.
The silence error also appears in the opposite direction: sounds that are present but too quiet to register as feedback. A footstep at −30 dB under a music bed is functionally silent. The equal loudness guide covers the perceptual matching; the beginner check is to mute the music and confirm that the feedback sounds are still clearly audible, then bring the music back and confirm they have not disappeared.
The seven errors above are not independent. The machine gun effect and the missing voice limit are both consequences of treating a sound as a single file rather than a system. The reverb and loudness errors are both consequences of treating a sound as a finished object rather than a layer in a mix. The timing and silence errors are both consequences of treating audio as decoration rather than feedback.
A beginner game does not need a professional sound design pass to avoid all seven. It needs five variants instead of one, a voice cap on the loudest sound, a reverb send instead of baked tails, a mix where SFX sits above music, headroom on the master bus, sound events on animation frames, and a confirm sound on every button. Those are not aesthetic decisions. They are the minimum that keeps the audio from reading as the reason the game feels wrong.