Delay one side of a mono signal by a few milliseconds and the ear stops hearing two copies. It hears one sound, wider, leaning toward the side that arrived first.
The precedence effect
Helmut Haas measured it in 1951: when the same sound reaches the ears from two directions within roughly 35 ms, the listener localises it to the first arrival and hears the second as part of the same event, adding width and body rather than an echo. That is the whole mechanism here. One channel carries the source; the other carries the same source a few milliseconds late.
| Delay | What you hear |
|---|---|
| 1 to 5 ms | the image shifts toward the undelayed side |
| 10 to 25 ms | a wide image that still reads as one source |
| 30 to 40 ms | width with a hint of slapback starting to separate |
Mono files
Most voice memos, phone recordings and dictation apps write a 1-channel file. This page opens such a file as two equal channels, so the delay has a second side to land on, and the export is a stereo file in the same format. A stereo file whose two channels are identical (dual mono) is treated the same way with no conversion needed.
Source
Sum to mono first averages both channels into one signal and builds the stereo image from that. Use it for mono and dual-mono material.
Delay the channel as it is keeps the existing left and right and delays only the chosen side. Use it on a real stereo recording when you want to push its image further toward one side.
Mono compatibility
Add the two sides back together and the delay becomes a comb filter. With a 15 ms delay the first notch falls at 33 Hz and the notches repeat every 67 Hz above it, which thins out a bass line badly on a phone speaker or a club system summed to mono.
Keep bass centred below sets a crossover. Below it both sides carry the identical, undelayed signal, so the low end sums cleanly. 150 Hz covers kick drums and bass guitar. Raise it toward 300 Hz for a full mix, or set it to 0 to delay the whole band.
Amount
Amount blends the delayed side between the undelayed source (0%) and the fully delayed one (100%). Below about 50% the result is a subtle widening that holds up better in mono.