The pre-delay trick that made 90s Mandopop vocals feel endless

I kept chasing that vocal sound. The one where the singer sounds like they're standing in a cathedral but you can still hear every breath. A-Mei does it. Jeff Chang does it. Basically every Mandopop ballad from the late 90s does it.

For months I was tweaking reverb tails. Longer decay, shorter decay, more high-frequency damping, less. Nothing worked. The vocals either sounded washed out or too dry — never that in-between where the voice feels open and spacious but still intimate.

The thing I was ignoring

Pre-delay. That's it. The gap between the dry vocal and when the reverb starts.

I'd always left it at the default — usually 0ms or whatever the plugin ships with. I didn't think it mattered. But when I started A/B-ing my covers against the reference tracks, the difference was obvious: the originals had a gap. The reverb didn't start immediately. There was a breath of silence.

What the pre-delay actually does

When you set pre-delay to something like 60–100ms, the dry vocal plays completely clean for that window. Then the reverb kicks in. Your ear hears the clarity of the close-mic'd voice first, and the space second.

Without pre-delay, the reverb smears into the transients. The "s" sounds get blurry. The breath before a phrase disappears into wash. With it, you get both: presence and depth.

The range I kept coming back to

After testing on probably 30 different vocal takes across different songs:

• 60–80ms for intimate ballads (think Jeff Chang's lower register)
• 80–120ms for bigger, more dramatic vocal moments (A-Mei belting)
• 40–60ms for faster-paced songs where you still want some space

The reverb itself matters less than I expected. A basic hall or plate with the right pre-delay sounds better than an expensive convolution reverb with no gap.

The second thing: decay time

Once I got the pre-delay right, I realized my decay times were too long. The 90s Mandopop records I was referencing used shorter tails than I expected — around 1.5–2.5 seconds for most ballads. Not the 4+ seconds I was using.

The combination of generous pre-delay and moderate decay creates that effect: the voice floats, but it doesn't drown.

What I do now

My default vocal reverb chain for Mandopop-style ballads:

• Pre-delay: 80ms
• Decay: 2.0s
• High-frequency damping: moderate (rolling off above 6–8kHz)
• Mix: 18–25% wet

I start there and adjust. The pre-delay is the thing I change last — it's usually right in that range.

---

This one took me embarrassingly long to figure out. The information is out there if you search for it, but none of the tutorials I watched connected it specifically to why old Mandopop vocals sound the way they do. It's not complicated. It's just a parameter I'd been ignoring.