Music and sound effects

The Music and Sound FX pages — two of the buttons in the Voices list’s title row — make audio from a description: describe it once, hear several takes side by side, keep the one that fits. Like voice-over, nothing here appears on a board or a node — every take is a file in Media, under Audio / Music or Audio / Sound FX, from the moment it is made.

The two pages work the same way. Sound FX is tuned for short sounds.

To hear a take in Play, put it in an Asset field on a node — the player plays it once when the story reaches that node (Sound on a node), and a Unity or Unreal game receives it as that step’s sound effect. A music track plays the same way: once, not looped, and not carried on to the next node.

Make a batch of takes

1. Open Voices and choose Music or Sound FX.
2. Pick an engine. Each row in the picker says what that engine is good for, and the controls below the description are that engine’s own — a control it does not take is not shown.
3. Describe it in Describe the music or Describe the sound — A heavy door sliding open on the station…. With the box empty, Start from offers a few examples.
4. Set Takes — how many to make from this one description.
5. Press Generate N takes, or Ctrl+Enter. The button shows what it will cost.

The takes arrive as a batch, each as it is ready. Under the button, the caption says whether the engine makes several takes per call or charges per take.

Improve the description

Improve rewrites your description in the form the chosen engine works best with. It generates no audio, and spends a small amount of credits for the rewrite.

The box is rewritten in place, with a line under it:

Link	What it does
Show yours · Hide yours	Shows your original words under the box
Back to yours	Puts your original back

If you switch engine afterwards, the line reads Written for «engine» and the button becomes Rewrite for «engine». A second rewrite always starts from your own words, not from the last rewrite.

The engine’s own controls

Which of these appear depends on the engine.

Control	What it does
Lyrics	Words to sing, with the parts marked — [Verse], [Chorus], [Bridge]. Leave it empty and the engine writes them from the description. Hidden while Instrumental is on.
Instrumental	No vocals. On an engine with no switch for it, it is said in the description for you, and the caption says so.
Keep out	What the take must not contain — music, voices
Length	A slider in seconds. Let it decide hands the length to the engine; where that is billed at a fixed length, the caption says how long.
Loops without a seam	A take that can repeat cleanly
Takes differ	From a lot to hardly — exactly as written: how far the takes of one batch stray from the description

Compare, rename and delete

Each take has a play button and a waveform — click the waveform to play from that point. Its name is on its row; click the name to rename it, which renames the Media file.

On a take	What it does
More like take N	A new batch from the same description and settings
Delete	Deletes the take’s Media file, with Undo on the notice that follows

On a batch	What it does
Use this description again	Puts the batch’s description and settings back in the box, to edit and run again
Stop	Stops the takes that have not started yet
Delete	Deletes every take in the batch, again with Undo

There is no separate “keep” step: every take is already saved. Delete the ones you don’t want.

A take that fails says so on its row — Take 2 failed, with the reason — is not charged, and offers Try again.

Layout

Feed and Workbench, in the page’s header, choose the layout; your choice is remembered in this browser.

Layout	How it reads
Feed	The batches read top to bottom, newest at the bottom, with the description docked underneath them
Workbench	The description on the left, the batches on the right, newest first

Plans and credits

Music and sound effects are on every plan; each take spends credits, as described on the Voices overview.