Auto sounds live or die on timing, perspective, and how much road noise you allow. When you prompt or choose a clip, decide what the viewer should notice (engine character vs tire texture), where the mic is (inside cabin, curbside, far across the street), and whether the tail needs to end cleanly for a cut. Use the tips below to get believable drivebys, starts, and skids without fighting your mix.
Name the action and the vehicle
Start with a clear action word (start, idle, rev, accelerate, pass-by, brake, skid) and a vehicle clue (compact, SUV, sports sedan, truck). This helps control the transient shape and low-end weight so the clip feels right for the picture.
- For sharp sync, include words like "tight attack" or "instant ignition"
- For heavier tone, mention "deep low-end" or "truck-like rumble"
- For clean foley moments, specify "no traffic bed" or "minimal background"
Set distance, movement, and stereo
Driving scenes often need motion: approach, pass, and recede. Add distance and camera position to get realistic doppler and stereo width, especially for curbside shots and fast pans.
- Close pass-by: ask for "wide stereo sweep" and "strong doppler"
- Distant street: ask for "narrow stereo" and "soft transient"
- Static shot: choose "steady idle" with "low movement"
Control texture: tires, wind, and reflections
Decide what texture should be present besides the engine. Tire hiss and wind can add realism, but too much will mask dialogue. Reflections (garage, alley, tunnel) change the tail and perceived size.
- Wet road: include "spray hiss" and a slightly darker tone
- Garage start: include "short reflections" and a tighter room tail
- Tunnel drive: request "fluttery reflections" and a longer decay
Avoid prompts that sound fake
Some prompts lead to unusable clips: overly loud distortion, random horns, or background crowds. Keep instructions specific and limit extra elements unless the scene demands them.
- Avoid stacking too many events in 5 seconds unless you need a montage hit
- If you need clean SFX, specify "no sirens, no voices, no music"
- If the clip is for layering, ask for "controlled noise floor" and "no clipping"
