Do not end the scene at the visible wall
Even when the camera shows a small kitchen, the scene does not have to stay inside those four walls. A brief low rumble from an elevator to the left, television thinned by a door, and a motorbike that turns two corners before reaching the window let the audience draw a corridor, a living room, and the width of an alley that never appear on screen. Sound is not simply enriching the background here. It is the invisible floor plan of the set.
The goal is not to add many sources. The scene needs clues that separate one distance from another. A nearby refrigerator motor can keep a dry low-mid hum and short vibration; footsteps at the end of a corridor should carry more reflection than direct sound. If every distance is flattened by the same reverb and brightness, the track count rises while the space becomes smaller. Three precise positions build a wider scene than ten vague ambiences.
That is also why recording only the sound of a quiet room is not enough when collecting room tone. Listen to what changes when a door opens, how the low end differs near the window and at the center of the room, and which wall returns a distant machine first. The identity of a place lives more clearly in the way its boundaries alter sound than in a fixed noise floor.
Place the source before choosing the emotion
Imagine glass breaking outside the frame. Boosting the high end and extending the tail can make the event bigger, but it can also blur where it happened. Set the direction of the first reflection, the delay between direct sound and reflection, and the frequencies lost through a wall, and the audience knows where the event is before turning its head. Surprise comes afterward. When position is believable, a small sound can become a threat; when position is vague, even a loud one floats like an effect.
An offscreen source must follow the same spatial rules as the character on screen. If a character walks from a bathroom into a living room, music outside that room should change in low end and early reflections as it crosses the doorway. The camera can remain still while changing occlusion and exposure produces motion inside the scene. Those shifts in spectrum and reflection often sell real movement better than simple left-right panning. Because position narrows quickly on small speakers, do not put direction only in level differences; distribute it across attack, filtering, and reflection time as well.
Exaggeration comes after the position is established. Horror may leave a silence longer than life and an unnaturally low reflection. Comedy may delay the sound by half a beat so it collides with a reaction shot. The first question, however, is not “how frightening?” It is “behind which door, and how far away?” Emotion grows stronger when it breaks a spatial rule, but the audience must hear the rule before it can be broken.
Leave the next event inside an empty frame
Offscreen sound expands editing time as well as physical space. In a shot of someone sitting without speaking, a distant bus can stop, open its doors, and pull away. Nothing changes in the image, yet the scene progresses. The audience begins to wait: did someone get off, and are they walking this way? The next action has been scheduled outside the frame without adding a cut. When that expectation is strong enough, music no longer has to keep explaining the emotion.
Attach every sound to an onscreen action and the scene becomes immediately legible, but it also closes quickly. A cup sounds when it lands, a door sounds when it opens, and an engine sounds only when the car appears; there is little left for the audience to infer. Let one or two sounds arrive before the image or remain after the cut, and an invisible bridge forms between scenes. The leading sound becomes a promise. The sound left behind becomes memory.
At the final pass, listening with the screen off can be as useful as checking the scene on mute. If you can roughly identify the entrance, the size of the room, the nearest object, and the event likely to happen next with your eyes closed, the world beyond the image is working. Every detail does not need to be literal. It only needs enough consistent evidence for listeners to imagine the same map. The frame is the boundary of the image; it does not have to be the boundary of the scene.