Every game we have passed through this desk, measured the same way: quiet dialogue against loud incident, at one fixed reference level, on one pair of headphones. The median gap is six decibels. In twenty-three of thirty-eight games it is wide enough that following a conversation and surviving a fight need different volume settings.
The complaint that a game's mix is muddy is universal and useless, because it describes a listener's impression without giving anyone else a way to check it. This survey exists to replace that impression with a figure, and the figure we chose is the gap between the quietest dialogue a player is expected to follow and the loudest incident a player is expected to survive.
The procedure is described on the approach page. In brief: set a comfortable level using a spoken passage, do not touch it again, then measure the loud material relative to that. If a player would have to change the volume to cope with both, the gap is too wide, and the number tells you by how much.
| Gap | Games | Share | Practical effect |
|---|---|---|---|
| Under 3 dB | 6 | 16% | Never touched the volume |
| 3–5 dB | 9 | 24% | Comfortable throughout |
| 6–8 dB | 13 | 34% | Occasional adjustment |
| 9–12 dB | 7 | 18% | Regular adjustment required |
| Over 12 dB | 3 | 8% | Two incompatible settings |
Fifteen games sit at five decibels or below and are comfortable across a whole session. Ten sit at nine or above, where a player must either accept inaudible dialogue or be startled regularly, and three are effectively asking for two different volume settings in the same game.
Thirty-one of the thirty-eight games offer a separate dialogue volume control, which sounds like a complete answer and is not. Raising dialogue in isolation frequently makes it sit oddly in front of the scene, and in nine games it introduced audible compression artefacts on the voice bus.
| Slider outcome | Games | Note |
|---|---|---|
| Fixes the gap cleanly | 14 | Voice remains natural in the scene |
| Fixes it, voice sits oddly forward | 8 | Legible but spatially wrong |
| Introduces artefacts | 9 | Audible pumping on the voice bus |
| No dialogue slider at all | 7 | Gap cannot be corrected by the player |
The seven games with no dialogue slider include two of the three worst gaps in the table, which is an unfortunate combination: the games most in need of player correction are among those least willing to permit it.
Six decibels is roughly the difference between comfortable and slightly too quiet. It is not catastrophic, and it is a strange median for an industry with excellent tools and decades of accumulated practice in exactly this problem.
A narrow gap is not automatically good, because it can be achieved by compressing everything into a narrow band. We therefore record dynamic range separately, from the softest passage in ordinary play to the loudest, and look for games that achieve a narrow dialogue gap while keeping range wide.
| Profile | Games | Assessment |
|---|---|---|
| Narrow gap, wide range | 8 | The target; skilful mixing |
| Narrow gap, narrow range | 7 | Comfortable but flat |
| Wide gap, wide range | 16 | Cinematic ambition, practical cost |
| Wide gap, narrow range | 7 | Loud and quiet in the wrong places |
Eight games manage the target profile. Sixteen are in the wide-and-wide group, which we read as a deliberate choice to sound like film, and film is mixed for a controlled room at a fixed level rather than a bedroom at eleven at night.
A cinema mix in a bedroom is not ambition, it is a mismatch between the room you mixed for and the room you shipped to.
Five strategy and management games were measured separately for how often a discrete alert interrupts the ambient bed, because the gap figure says nothing useful about a game where the problem is frequency rather than level.
| City or empire size | Best result | Median | Worst result |
|---|---|---|---|
| Early | 1 per 240 s | 1 per 95 s | 1 per 48 s |
| Moderate | 1 per 90 s | 1 per 34 s | 1 per 11 s |
| Large | 1 per 55 s | 1 per 19 s | 1 per 6 s |
One alert every six seconds is not a soundscape. Two of the five games scale alert frequency with importance rather than with the number of entities that could produce one, and those two are the only ones we could still hear the music in at large scale.
Keep the gap under five decibels while leaving the range wide, which eight games in this table demonstrate is achievable. Provide a dialogue slider that does not compress the voice bus when raised. Where a design genuinely requires a wide gap, say so somewhere a player will see before buying, and scale alerts by importance rather than by count.
