Audio Drama Script Format: Template and Examples
Audio drama script format: scene headings, NAME: dialogue, (directions) and SFX cues. Copy the template, read original examples and hear them performed.
Last updated: 2026-10-26
Audio drama script format puts one speaker on each line, NAME: dialogue, with numbered scene headings, short delivery notes in round brackets and sound or music cues on lines of their own. It is close to a stage play and far from a screenplay: there is nothing to see, so every place, entrance and action has to reach the listener as sound or speech. Below are the five elements, a copy-paste template, two original example scripts with their openings performed, a Poe story turned from prose into script, and how each kind of line behaves in our audio drama studio.
The five elements of an audio drama script
| Element | How to write it | Example | In the VoiceSmiths studio |
|---|---|---|---|
| Scene heading | Number, place, time, in capitals | SCENE 1. A FARMHOUSE BEDROOM. NIGHT. | Read aloud by the narrator. Leave it out of the paste if it belongs on paper only |
| Narration | A NARRATOR: line | NARRATOR: The Harlows moved in a month ago. | Narrator voice. NARRATOR, NARRATION, V.O. and SCENE cues all go to the narrator |
| Dialogue | Name, colon, the line | TOM: Go back to sleep. | Each new name becomes a character with a voice of its own |
| Direction | A word or two in brackets, before the colon | JUNE (whispering): Did you hear that? | Used as the line's emotion and never spoken; 24 characters at most |
| Sound and music cues | SFX: or MUSIC: on a line of its own | SFX: Three slow footsteps overhead. | Not performed. The studio would cast "SFX" as a character and read the cue, so remove these lines from the paste and keep them for the sound pass |
Three rules make a script read cleanly, for actors and for the studio alike:
- Spell every name the same way, every time.
JUNEandJuneare two different characters to a parser. BBC Writersroom's template makes the same point: the character name "should remain consistent throughout the entire script." - Keep names short. Up to three words starting with a capital letter:
JUNE,OLD MAN,DR. MORROW. - One speaker per line. If two people speak at once, give each a line.
The studio switches to script mode when at least 40% of the lines you paste start with a cue (NAME:); in a script, any line without a cue is read by the narrator.
Audio drama script vs screenplay
| Screenplay | Audio drama script | |
|---|---|---|
| Scene heading | INT. KITCHEN - NIGHT | SCENE 1. KITCHEN. NIGHT., and the place must also be heard |
| Action lines | Describe what the camera sees | None; sound cues and dialogue do the work |
| Character names | Centred above the dialogue | In capitals, then a colon, dialogue on the same line |
| Parentheticals | (beat), (sotto) | Delivery and microphone: (OFF), (CLOSE), (V.O.), (LOW) |
| Transitions | CUT TO: | A sound, a music sting or a narrator line |
Copy-paste audio drama script template
TITLE: [Title]
EPISODE 1: [Episode title]
SCENE 1. [PLACE]. [TIME].
SFX: [One sound that tells us where we are.]
NARRATOR: [One or two sentences, only what the sound cannot say.]
CHARACTER A: [First line. Name the person they are talking to.]
CHARACTER B (whispering): [Reply. One emotion per direction.]
CHARACTER A: [Line.]
MUSIC: [Sting or transition.]
SCENE 2. [PLACE]. [TIME].
[...]
END
This is the full working script, for you and for whoever does the sound. Before pasting it into the studio, make a voice copy: delete the TITLE, EPISODE, SFX, MUSIC and END lines, and keep the scene headings only if you want them read aloud.
The BBC radio layout
BBC Writersroom publishes a radio drama template in "scene style", the BBC's house format for radio drama, plays and non-audience comedy. Its conventions:
- Scenes are numbered:
SCENE 1.,SCENE 2. - Technical directions for sound and music are written in underlined capitals, set apart from the dialogue, and used sparingly: the template asks you to "never use superfluous novelistic text."
- Character names are in capitals with a colon, and the dialogue starts on the same line.
- Delivery notes go in capitals in round brackets inside the line: (BEAT), (PAUSE), (OFF) for away from the microphone, (V.O.), (D) for a distorted voice on a phone or radio, (LOW), (CLOSE).
- The script ends with
END.
To paste a BBC-style script into our studio, move each emotional note in front of the colon and write it as a feeling: JUNE (LOW): Did you hear that? becomes JUNE (whispering): Did you hear that?. Microphone notes such as (OFF) and (CLOSE) are for a recording studio and have no effect on AI voices; for a voice on a phone line, use the phone voice effect in the voice studio.
Example 1: a horror cold open
An original scene of about a minute, written for this page. Three voices, four sound cues.
THE SECOND FLOOR
by VoiceSmiths
SCENE 1. A GROUND-FLOOR BEDROOM IN AN OLD FARMHOUSE. JUST AFTER MIDNIGHT.
SFX: Rain against the window. A clock ticks.
NARRATOR: The Harlows moved in a month ago. They have never been upstairs. The agent said the second floor was sealed in 1974, and nobody asked why.
JUNE (whispering): Tom. Tom, wake up. Did you hear that?
TOM: It's the house settling. Go back to sleep.
JUNE (nervous): Houses don't settle in footsteps.
SFX: Three slow footsteps, directly overhead.
TOM: ...Okay. That was footsteps.
JUNE: The stairs are boarded up, Tom. Nobody can get up there.
TOM (whispering): Then who is walking around up there, June?
SFX: Above them, a door opens.
MUSIC: One low string note, held under the next scene.
The opening, voiced line by line on Eleven v4 with three of our designed voices: British Butler as the narrator, Sweet Girl (EN) as June, Aussie Surfer as Tom. The directions became tags; the sound cues were left for the sound pass:
[whispers] Tom. Tom, wake up. Did you hear that?
It's the house settling. Go back to sleep.
[nervous] Houses don't settle in footsteps.And this is the copy you would paste into the studio, with the sound and music lines taken out:
NARRATOR: The Harlows moved in a month ago. They have never been upstairs. The agent said the second floor was sealed in 1974, and nobody asked why.
JUNE (whispering): Tom. Tom, wake up. Did you hear that?
TOM: It's the house settling. Go back to sleep.
JUNE (nervous): Houses don't settle in footsteps.
TOM: ...Okay. That was footsteps.
JUNE: The stairs are boarded up, Tom. Nobody can get up there.
TOM (whispering): Then who is walking around up there, June?
Notice what the dialogue does that a screenplay would leave to the camera: both names are said in the first two lines, the footsteps are named out loud, and the boarded-up stairs are explained in speech because nobody can see them. For a full episode in the same genre, here is an original horror short from our theater, about four and a half minutes with six voices:
Example 2: a two-hander comedy
Two voices are the easiest audio drama to follow, because every line answers the one before it. Another original scene:
SCENE 1. A BUSY CAFE. SATURDAY MORNING.
SFX: Cups, chatter, an espresso machine.
MAYA: Excuse me, is this seat taken?
LEO: It's reserved. For my date. Who is forty minutes late.
MAYA (laughing): Mine's forty minutes late too.
LEO: Let me guess. Tall, says "no worries" a lot, sent you a photo with a dog?
MAYA: ...Yes.
LEO: Same dog.
SFX: Two phones buzz on the table, one after the other.
MAYA (surprised): He just cancelled on me.
LEO: He just cancelled on me too.
MAYA: Well. The seat's free now.
MUSIC: Light piano, in and under.
The first four lines, performed by Sweet Girl (EN) and Tech Founder:
[laughing] Mine's forty minutes late too.
Let me guess. Tall, says "no worries" a lot, sent you a photo with a dog?The joke lands on two words, "Same dog", which is how comedy works on audio: short lines, a pause, a payoff. For pacing over a full episode, here is a romantic comedy from the theater:
Example 3: turning prose into a script
Adapting a story mostly means splitting the narration from the speech. Here is a passage from Edgar Allan Poe's The Tell-Tale Heart (1843, public domain), as Poe wrote it:
I had my head in, and was about to open the lantern, when my thumb slipped upon the tin fastening, and the old man sprang up in bed, crying out—"Who's there?"
I kept quite still and said nothing. For a whole hour I did not move a muscle, and in the meantime I did not hear him lie down. […] Presently I heard a slight groan, and I knew it was the groan of mortal terror. […] He had been saying to himself—"It is nothing but the wind in the chimney—it is only a mouse crossing the floor," or "It is merely a cricket which has made a single chirp."
The same passage in script format, as it was performed (abridged, like the rest of the episode) in our theater:
NARRATOR: I had my head in, and was about to open the lantern, when my thumb slipped upon the tin fastening, and the old man sprang up in bed, crying out—
OLD MAN: Who's there?
NARRATOR: I kept quite still and said nothing. For a whole hour I did not move a muscle, and in the meantime I did not hear him lie down.
NARRATOR: Presently I heard a slight groan, and I knew it was the groan of mortal terror. He had been saying to himself—
OLD MAN (whispering): It is nothing but the wind in the chimney—it is only a mouse crossing the floor,
NARRATOR: or
OLD MAN: It is merely a cricket which has made a single chirp.
Two lessons from this one. First, Poe's narrator is the killer telling his own story, so in a first-person piece the narrator is a character: cast him like one, not as a neutral reader. Second, keep the author's words: "crying out—" stays with the narrator, and the old man says only what is inside the quotation marks.
You do not have to convert prose for our studio: paste the story as Poe wrote it and it works out who speaks each quotation. Converting pays off when the speakers are ambiguous or when actors will read it. More Poe, performed in full: Edgar Allan Poe full-cast audio.
Writing for the ear: eight rules
- Say names out loud, especially in the first minute. The BBC template puts it plainly: "The only means of establishing a character's presence is to have them speak or be referred to by name."
- Keep scenes to four speaking voices or fewer. "If there are too many characters in a scene, the listener will lose track."
- Set the place with one sound and one line, then move on.
- Give each character a way of talking: sentence length, a pet phrase, how formal they are. Casting voices of different ages and genders helps too.
- Keep lines short, one thought each. Long speeches belong to a narrator or a monologue.
- Announce entrances and exits with a line ("Oh, you're back.") or a sound (a door).
- Use directions sparingly: one at each emotional turn, not on every line.
- End every scene on a turn: a question, a reveal or a sound.
How long should an audio drama script be?
A page of script is about a minute of audio, and English drama runs at roughly 150 words a minute. Our own numbers: across the 172 English episodes in the theater, 1,000 characters of script came to about 71 seconds, or roughly 845 characters a minute.
| Episode length | Words | Characters | Credits to render on VoiceSmiths |
|---|---|---|---|
| 5 minutes | ~750 | ~4,200 | ~300 |
| 10 minutes | ~1,500 | ~8,500 | ~600 |
| 15 minutes | ~2,250 | ~12,700 | ~890 |
| 30 minutes | ~4,500 | ~25,000 | ~1,750 |
The studio reads up to 8,000 characters per paste, about nine and a half minutes of English, so split a longer episode at a scene break.
From script to finished audio
- Paste the voice copy into the audio drama studio. Finding the cast needs credits in your account but uses none.
- Check the cast. Every name gets a voice; tap a face to hear it and pick another if it doesn't fit. For a series, add each episode to the same series so every character keeps their voice.
- Check the lines. Each line shows its speaker and emotion, and you can change either from a list.
- Render. You get an MP3, SRT subtitles and the script; a failed render refunds its credits. By default the file ends with a spoken "This audio was generated by AI", which you can switch off before rendering.
- Do the sound pass. Make each SFX and MUSIC cue with the AI sound effects generator and place it against the subtitle timings in your editor.
Rendering costs 70 credits per 1,000 characters, so a 10-minute English episode is about 600 credits: roughly $12.50–19.90 at monthly-plan list prices ($5.00–7.90 in the first month at the new-user price); see pricing. The studio reads Chinese and English scripts; a finished episode can then be adapted into Japanese, Korean, Spanish, Portuguese, French, German or Indonesian with the same cast. For more on AI-performed audio plays, see AI radio drama.
Paste your own scene below to see how it is read and cast:
Generating needs credits · trial pack from $3.90 · hear finished dramas in the theater first
FAQ
What is the format of an audio drama script? One speaker per line: the name in capitals, a colon, then the dialogue, as in JUNE: Did you hear that? Scenes start with a numbered heading, sound and music cues sit on lines of their own (SFX:, MUSIC:), and short delivery notes go in round brackets. There are no camera directions: everything the audience needs has to be heard.
How long is one page of an audio drama script? As a rule of thumb, about a minute of audio. English drama runs at roughly 150 words a minute; across the 172 English episodes in our theater, 1,000 characters of script came to about 71 seconds.
Do sound effects go in the script? Yes, on lines of their own, so whoever does the sound can find every cue. Our studio renders voices only, so take the SFX and MUSIC lines out of the copy you paste and use them as the list for your sound pass.
Can an AI read stage directions like (angrily)? In the VoiceSmiths studio a short direction in brackets between the name and the colon, as in JOHN (angrily):, is used as that line's emotion and is not spoken. A direction written after the colon stays part of the line and may be read out, so move it in front of the colon.
Is there a standard BBC radio script format? Yes. BBC Writersroom publishes a radio drama template in its house "scene style": numbered scenes, sound and music directions in underlined capitals, character names in capitals with a colon, and delivery notes such as (OFF), (V.O.) and (CLOSE) in brackets. The plain format on this page keeps that structure and pastes cleanly into tools.
Do I have to turn a story into script format? Not for our studio: prose with quoted dialogue works too, and the AI works out who says each line. Script format pays off when the speakers are ambiguous, when you want to fix exactly who says what, or when real actors will read it.