A sound effect that depends on sub-bass for its identity does not exist on a phone speaker. That is not a mixing preference or a matter of taste. Laptop and tablet playback systems attenuate frequencies below roughly 200 Hz and above 15 kHz; a phone speaker is designed primarily for speech and is most efficient in the 700 Hz–2.5 kHz range; and many of these devices sum stereo to mono, which means an element that relies on wide stereo placement can partially or entirely cancel.
The practical consequence is that a sound designed and approved on studio monitors can lose its defining characteristic on the device most players will actually use. The fix is not to abandon low end. It is to decide where the sound's identity lives, and to make sure that identity survives when the bottom two octaves are removed and the stereo field collapses. The table below shows the three device constraints that matter, and what each one does to a sound effect that is designed without accounting for it.
| Device constraint | What it removes | What happens if the sound's identity is there |
|---|---|---|
| Roll-off below ~200 Hz | Sub-bass and low body | Impact reads as a tick or a click; weight disappears |
| Roll-off above ~15 kHz | Air and high transient detail | Sound becomes dull; attack loses definition |
| Mono summing | Out-of-phase stereo content | Layer disappears or partially cancels when summed |
The rest of this guide is about moving the identity to a band those devices can reproduce, and testing whether the move worked. The mobile file size and performance guide covers the encoding and memory side of the same problem; this article is about the acoustic side, which is a separate decision.
Put the Identity in the Speech Band, Not the Sub
A phone speaker is built around voice intelligibility. That is why it is most efficient in the 700 Hz–2.5 kHz region. A game sound effect that carries its recognisable character in that band will remain present on a phone, a laptop, and a pair of cheap earbuds, because all of those devices are designed to reproduce speech.
This does not mean the sound should sound like speech. It means the layer that tells the player what event just happened should have energy in the midrange. A weapon impact whose identity is a low thud will vanish on a phone. A weapon impact whose identity is a short midrange crack, with the low thud layered underneath as support, will still read as an impact when the thud is gone. The guide to layering game sound effects without muddy mixes covers how to split layers by frequency so the midrange carrier is not competing with the low-end support for space.
For a generator-based workflow, this translates to a specific decision: before adding a sub layer or a low-frequency oscillator, render the sound without it and listen on a phone speaker or a mono laptop speaker. If the event is unrecognisable, the identity was in the wrong place. Move it up before you polish the low end. The explosion sound effect guide shows how noise shaping can carry impact weight through midrange energy rather than relying solely on sub-bass, and the same principle applies to any impact-based sound.
Use Upper Harmonics to Imply the Low End That Is Missing
A small speaker cannot move enough air to reproduce a 60 Hz fundamental at a meaningful level. It can, however, reproduce the harmonics above that fundamental. The auditory system reconstructs the missing fundamental from those harmonics — a phenomenon often called the missing fundamental effect — and the listener perceives a low end that the speaker is not actually producing.
The technique is to add or emphasize the harmonics of the low-frequency layer rather than trying to force the fundamental through the speaker. Tracy Bush, who designed sound for R2-D2 and other embedded devices, described this approach in an interview with A Sound Effect: introducing carefully controlled upper harmonics so the auditory system infers the presence of frequencies the speaker cannot reproduce, rather than attempting to overcome the hardware's physical limits. The result is not identical to full-range playback, but it is a convincing impression of weight.
In a browser-based generator, this can be approximated by adding a layer at one or two octaves above the intended fundamental. A 60 Hz sub layer, for example, can be paired with a 120 Hz and a 240 Hz layer at lower amplitude. On a full-range system, the sub dominates and the harmonics add definition. On a phone speaker, the sub disappears and the harmonics carry the perceived weight. The synthesizer guide for laser and sci-fi sounds covers how oscillator stacking affects perceived brightness and weight, and the same stacking logic applies here in reverse: you are stacking upward to compensate for what the device removes.
Check Mono Before You Check Anything Else
Phone speakers and many laptop speakers play in mono or close to it. A stereo mix that relies on a wide element — a panned layer, a stereo widener, a phase-inverted duplicate — can lose that element entirely when summed to mono, or change its character in a way that was never audible in stereo.
The check is straightforward: sum the sound effect to mono and listen. If an element disappears or becomes significantly quieter, it was relying on phase relationships that the mono sum cancels. The UI sound design guide makes the same point about confirmation and cancel sounds: they need to remain distinguishable when played through a single small speaker, not only on headphones. The fix is usually to keep the identity-carrying layer centred and use stereo only for decorative width that is not carrying information.
A practical rule for short game sounds: if an element is important enough that the player needs to identify the event, it should be present at roughly equal level in both channels. Stereo can be used for spatial placement of the source in 3D audio, but the dry, identity-carrying layer should be mono-safe. The 3D positional audio guide covers the cases where spatial placement is carrying information, and where the mono-safe rule does not apply.
The Device Test That Catches Translation Failures
Studio monitors and headphones do not reveal the problems that phone speakers reveal. A mix that sounds balanced on both can still lose its key elements on a phone, because the phone is reproducing a different frequency range entirely.
The test is not “does it sound good on a phone.” It is “does the event still read on a phone.” Play the game or the sound on the actual device, at a normal volume, in a normal room — not a treated studio, not with headphones. The question is whether the player can tell what happened. A footstep should still sound like a footstep on a specific surface; a hit should still sound like a hit; a coin should still sound like a coin. If any of those become ambiguous when the low end is removed, the identity is in the wrong band.
For a sound designer without a phone-testing setup, a mono-summed laptop speaker at low volume is a reasonable approximation of the worst case. The Nintendo Switch sound design guide covers the handheld speaker case in more detail, including the volume and frequency constraints that are similar to phone playback. The principle is the same across devices: design for the speaker, not for the monitor, and treat the monitor as the exception rather than the target.
Two Versions, One Event
A sound effect that survives on a laptop and phone speaker is not a lesser sound. It is a sound whose identity was placed deliberately in the band those devices can reproduce, with the low end and stereo width treated as enhancement rather than foundation.
The most reliable way to ship that is to author two versions from the same starting point: a midrange-forward version for the default mix, and a full-range version for headphones or console playback where the low end has somewhere to go. The difference between them is usually a layer decision, not a different sound — the same midrange crack, with or without the sub layer underneath. That keeps the event recognisable in both contexts and avoids the common outcome where a sound is tuned for one playback system and fails silently on the other.