3 July 2026 | 6 min read
Dolby Atmos explained: beds, objects and why your stereo mix still matters
Atmos is not more speakers, it is metadata. The mix describes where sounds are and the room decides how to play them.
Traditional surround assigns audio to channels: this sound goes to the left surround speaker. Dolby Atmos assigns audio to positions: this sound is here in the room, work out which speakers should produce it. That single change is what everything else follows from.
Beds and objects
- A bed is a conventional channel based layer, usually 7.1.2, carrying music, ambience and anything that does not need to move.
- An object is a mono audio stream plus positional metadata over time, up to 118 of them in a cinema mix.
- The renderer combines beds and objects for whatever playback layout exists, from a 64 speaker cinema to a phone.
What this changes for production sound
Very little on set, and quite a lot at ingest. Objects are mono. A mix that will end up in Atmos wants clean, separate, well named mono sources rather than pre-summed stereo stems. A poly WAV split into named mono tracks is exactly the right starting material.
Deliverables
A typical package includes the Atmos master file, a 5.1 fold down, a stereo fold down and often a binaural render for headphones. The fold downs are usually generated by the renderer, but they still need checking, because dialogue that sits perfectly in an immersive field can get lost in a LoRo stereo reduction.
Checking a fold down
The fastest sanity check is to make a manual LoRo downmix of your surround stems and listen on a laptop speaker. Standard practice puts L and R at unity, the centre and surrounds around minus 3 dB, and the LFE excluded. If the dialogue disappears in that version, it will disappear for most of your audience.
You can do that reduction here in the browser, with a fader per channel, and export the result without uploading anything.