This keyword is usually about a spoken command, so the best result depends on who is speaking, how far the camera feels, and whether the line is heard directly, through a door, or over comms. Use the library for quick pulls, or prompt the AI to regenerate with the exact emotion, pacing, and space you need.
Pick the emotion and intent first
A scared warning reads differently than an aggressive order. Decide if the line should feel protective, threatening, embarrassed, or comedic, then choose a take with matching intensity and vocal texture.
- For danger: panicked yell with breaths and a slightly longer decay
- For intimidation: close, hard shout with strong consonant attack
- For comedy: dry, quick delivery with minimal tail
Match distance and obstruction
Distance changes the tone and clarity. If the character is offscreen or behind something, a clean close-mic shout will feel fake. Choose a take that already suggests space, or regenerate with distance cues in the prompt.
- Close: crisp transient, low room tone, centered voice
- Distant: rolled-off highs, more air, softer edges on consonants
- Behind a door: muffled top end with boxy resonance and a short tail
Choose the right tail and room tone
The tail is what sells the location and also what can ruin an edit if it overlaps the next line. Short tails cut cleanly. Longer tails feel cinematic but need a gap in the dialogue.
- Hallway or stairwell: audible early reflections and a longer decay
- Small room: tight reflections with a compact tail
- Cut-to-black: controlled tail that ends decisively without ringing
When to use radio or PA filtering
If the command is coming through comms, do not use a full-range vocal. A band-limited, slightly noisy texture signals radio realism and helps the line sit under loud action sounds.
- Radio: bandpass tone, light static, compressed dynamics
- Intercom: slightly metallic resonance with limited low end
- Avoid: overly wide stereo or lush reverb that breaks the comms illusion
