Align Delay: small timing offsets, large stereo changes

Listen to a timing change without a new echo

This example delays the right channel by 1.5 ms and leaves the left at zero. The original synthetic stereo source contains shared tones and opposite-polarity side tones; its shared component drops in level halfway through. These are sustained signals, so listen for a change in the image and the way the tones combine, not for a second drum hit.

Keep Delay Mode on Time and Link off. During playback, return Delay R from 1.5 ms to 0 ms. Let it settle, then compare the same section again at 1.5 ms. The intended cue is a change in the relationship between the ears, particularly in the side-heavy second half. The source already has deliberate stereo differences: zero restores its original relationship, not a promise of a perfectly centred image.

Try 3 ms next, then return to zero. A larger delay does not necessarily sound proportionally wider. Each frequency travels through a different fraction of its cycle during the same time offset. Some relationships can reinforce while others cancel when the channels are combined. Headphones make the stereo comparison accessible, but a mono check in your working mix is still necessary before keeping the effect.

Link changes the question

Set Delay L to 20 ms and enable Link. Both output channels now follow the left delay, regardless of the stored Delay R value. Their relative timing is preserved while the entire stereo signal arrives later. On an isolated sustained sound, this common delay need not make an obvious tonal change; against a separate undelayed track it becomes a timing decision.

Move Delay L while linked, not Delay R. Turn Link off again: the right channel returns to its own stored amount. Set both delays to zero when you finish the comparison. Linking does not permanently copy the left value into the right setting, so an old offset can reappear when the link is released.

When there really are two arrivals to align

Imagine two related recordings of the same event, one arriving earlier than the other. Delay the earlier one toward the later one, starting with a measured transient difference where possible. Listen to the combined result, not just the soloed delayed track. An apparent gain in body may indicate better agreement, but confirm that attacks and other notes also benefit.

The device only adds positive delay, from 0 to 500 ms per channel. It cannot advance a late recording, detect transients, estimate the best offset, or correct frequency-dependent phase differences with one number. Separate tracks need appropriate instances or routing in your project; this single-source demonstration does not secretly provide a second microphone recording or an automatic alignment reference.

Three readouts, one stored delay

Time, Samples and Distance are display modes for the same millisecond delay. Switching mode does not change the sound. The Samples readout uses a nominal 48 kHz reference, so 1 ms is shown as 48 samples. The actual delay remains stored in time: do not treat that label as a measured sample count at every audio-context sample rate.

Distance uses 343 metres per second, giving about 0.34 m for 1 ms. That conversion can help interpret an offset, but it is not a measurement of your room, speaker placement or current speed of sound. Stay in Time for the listening exercise so the numbers describe exactly the delay being applied.

There is no dry/wet control, feedback loop or echo train. Every channel is simply shifted. Use a repeat delay for audible echoes and Utility for polarity and mono checks. If a small stereo offset sounds attractive, compare it in the full arrangement and in mono before calling it an improvement rather than merely a different image.