Your AI hosts can now laugh, sigh, whisper, pause, and react.
Now, with audio prompting, you can direct how each line in your script is spoken or add additional, non-verbal sound effects and reactions. Type / inside any speak block in the script editor to open a dropdown of preset cues, or type something custom.
Audio tags are organized into three main categories:
- Reactions like [laughs], [sighs], and [gasps]
- Emotions like [excited], [nervous], and [sarcastic]
- Delivery cues like [whispers], [pauses], and [dramatic].
For example, writing "I can't believe it [gasps] that's incredible! [excited]" will produce a line with a sharp intake of breath followed by an energetic, upbeat delivery. Each tag appears in inline in your script so you can see exactly how a line will play before you generate.
Generated scripts will now also occasionally include audio tags to improve the naturalness of your podcasts and conversations.
Audio prompting only affects what your listeners hear, keeping your captions, transcripts, and visual assets clean.
For the full picture of what actually makes an AI voice sound natural instead of robotic, script rhythm, tag choice, and picking the right voice for the job, see how to make an AI voice sound human.
For a worked example of using these cues as the delivery notes in a multi-host scene, see how to write a podcast script.