An audio object is the basic unit of object-based audio: a single sound — a voice, a passing car, a helicopter — bundled with metadata that describes its position (x, y, z), size, and level over time. Instead of being mixed into a fixed channel, the object stays separate until playback, when a renderer places it on the listener's actual speakers or headphones according to that metadata. Keeping sounds as discrete objects is what gives Dolby Atmos and MPEG-H their flexibility: objects can be positioned with height and precision, moved dynamically, individually adjusted (a dialogue object boosted for clarity), or swapped (a different language object). A typical Atmos mix combines a channel bed with up to a hundred or so such objects.
