Eleven prompts, eleven takes.
Nothing here is theoretical.
Every prompt below is shown exactly as written, next to the unedited audio it rendered. Copy one, open it in the playground, change a single cue, and hear what that cue does. That loop teaches faster than any tutorial.
New to the grammar? Read the cue sheet guide first.
Casting voices
Age, texture and dynamic between speakers, each in a phrase. Voices stay consistent for the whole take.
Breakfast table. A cheerful six year old girl begs to keep a stray cat, her firm but warm father resists, and her raspy grandmother takes the girl's side, laughing. Kitchen clatter and a kettle behind them.Three voices, each cast in half a sentence. The ages do the differentiation work.
A phone call. An angry middle aged customer complains fast and loud about a late delivery. A young female support agent stays calm, soft and professional, and slowly wins him over. Faint office hum.The emotional arc is written as an instruction ('slowly wins him over'), not as adjectives.
A gentle elderly female voice reads a bedtime story, slow and tender, almost a whisper: 'And so the little fox finally found its way home.' A soft music box plays far away.Register direction ('almost a whisper') controls intimacy better than the word 'soft'.
Directing energy and beats
Tempo, intensity and where the emphasis lands. Direct the delivery like a stage note.
Two excited male sports commentators call the final seconds of a football match. The older one is hoarse and rapid, the younger one screams as the winning goal lands. Distant stadium roar swells.The contrast between the two voices is specified, so energy reads as dialogue, not noise.
A deep, authoritative movie trailer narrator, slow and dramatic: 'This summer, one city. One chance. No way back.' Low cinematic hits punctuate each sentence.Short quoted lines with hard punctuation give the model beats to hit.
Placing music and ambience
Name the bed, then give it entrances and exits. The mix happens inside the render.
A two-host podcast cold open. An upbeat female host teases today's story, her male co-host reacts with a laugh and one skeptical question. A short synth intro sting fades under their banter.The sting gets an exit cue ('fades under their banter'), exactly like an edit note.
A 1940s radio drama cold open. A low, gravelly male narrator speaks over faint vinyl crackle and a slow, smoky jazz trumpet: "The city sleeps. Somewhere out there, a phone is ringing that should not be."Era plus medium ('1940s radio', 'vinyl crackle') sets tone, music style and diction at once.
Steady rain falling on a tin roof, distant soft thunder, and a lone melancholy cello playing a slow melody. No speech, just the scene.'No speech, just the scene' unlocks pure-ambience mode for loops and beds.
Building worlds
One sentence of place tunes everything at once: the voice, the props, the reverb.
A medieval tavern in a fantasy game. A gruff dwarven innkeeper greets the player: 'Back again, adventurer? The usual corner table is yours.' Fireplace crackle, a lute in the corner, mugs clinking.One sentence of world anchors the voice, the props and the room together.
Two street vendors argue playfully over the last crate of mangoes at a busy night market. Lively crowd chatter and sizzling food carts behind them. The older vendor laughs first and gives in.The background is told to stay behind the voices, keeping dialogue in front of the crowd.
An audiobook narrator, warm middle aged male, reads: 'The lighthouse keeper counted the ships every night. Tonight, one light too many burned on the horizon.' He gives the last sentence a chill. Faint waves and wind under the voice.One targeted direction on the final sentence beats ten adjectives up front.
Make the twelfth one yours
Steal the pattern that matches your project, swap the world and the cast, and render it free in the playground.