Matrix Surround
Matrix surround encodes spatial relationships into fewer channels for later surround decoding.
How can a two-channel recording produce center and surround outputs? Matrix surround deliberately combines spatial signals into a pair with recognizable level and phase relationships. A matching decoder uses those relationships to steer the presentation.
The transmitted pair is commonly labeled Lt/Rt, left total/right total. It can be heard as stereo as well as passed to a compatible matrix decoder.
Understand the cue before the decoder
A deliberately simplified mathematical model can illustrate the idea:
Lt = L + C + S
Rt = R + C − S
This is not the actual Dolby matrix. Real systems use defined coefficients and phase processing; the model only exposes the information relationships.
If only C is active, Lt and Rt match. If only S is active, they oppose each other. A decoder examining sum and difference can respond to those two cases:
| Model input | Lt + Rt | Lt − Rt |
|---|---|---|
| Center only, C | 2C | 0 |
| Surround only, S | 0 | 2S |
| Left only, L | L | L |
Matching and opposing relationships provide cues. Dolby's encoder documentation likewise specifies out-of-phase Lt/Rt outputs for surround-channel test signals, using its actual encoding process.
Why two signals cannot preserve every independent channel
In the model, arbitrary L, R, C, and S are four independent unknown signals described by only two transmitted signals. There is no unique algebraic solution for all possible mixtures. A center-only input and an appropriate matching front-left/right input can produce the same Lt/Rt.
A practical decoder uses analysis and steering to emphasize the likely dominant directions and manage the ambiguous contributions. This can produce an effective surround impression, especially for material prepared for the matrix. It does not provide the unrestricted channel separation of a discrete multichannel recording.
Later matrix systems use more sophisticated relationships and output arrangements. The specific encoding/decoder pairing determines which cues and outputs are intended.
Carry the intended stereo pair
Lo/Ro, left only/right only, describes an ordinary stereo downmix without that intentional surround-matrix treatment. Both Lo/Ro and Lt/Rt have two stored channels, but they ask different things of a surround decoder.
For an Lt/Rt soundtrack in an Emby file, preserving the pair's relative phase and level preserves its decoding cues. Taking only one side, independently altering phase, or making an unsuitable mono conversion can change those relationships.
Applying an upmixer to arbitrary stereo derives extra outputs from whatever relationships are present. Decoding an intentionally encoded matrix uses relationships the producer prepared for that purpose. A virtualizer can then take the resulting surround positions and reproduce their cues over headphones or fewer speakers, performing a subsequent acoustic-rendering job.