Manual Emotion Tags with Chatterbox FAL in LyricWinter

Chatterbox FAL emotion tags are short inline cues that ask the model to add a specific nonverbal performance beat. In LyricWinter, they are most useful for dialogue that needs a laugh, sigh, gasp, yawn, groan, cough, chuckle, or sniffle at a specific point in the line.^[1]

The Chatterbox API documents these tags directly on the text input for fal-ai/chatterbox/text-to-speech. The syntax uses angle brackets, such as <laugh> or <sigh>, not the square brackets used by Fish Audio in LyricWinter.^[2]

The important mental model is simple: LyricWinter detects speakers and assigns voices, but Chatterbox emotion tags are manual direction. If you want Chatterbox to perform a cue, put that cue in the text that becomes audio.

The Syntax

Place the tag exactly where the sound should happen. A tag at the beginning colors the opening of the line. A tag in the middle asks for the sound at that beat. Use lowercase tags with angle brackets.

Mira: <sigh> I knew the lock would recognize you.
Sela: Don't say that like it's normal. <gasp> It just blinked.
Bram: <groan> We are absolutely not opening the singing door.
Mira: <chuckle> You say that every time.

Chatterbox still uses the rest of the sentence, punctuation, and context. Tags are performance cues, not replacements for clear writing. Use them when the beat matters enough that the model should not have to infer it from prose alone.

Complete Chatterbox Tag Reference

The current Chatterbox FAL endpoint documents eight inline emotion or sound tags for text-to-speech. LyricWinter sends your line text through to Chatterbox, so these tags should be written directly into the line you want performed.^[2]

Where to Put Tags in LyricWinter

There are two practical workflows.

• Before detection: paste a script or story that already contains Chatterbox tags. This is fastest when you know the beats before generation.
• After detection: click a detected line, edit the line text, add the tag, confirm the edit, then generate or regenerate audio. This is better when you want to inspect speaker splits first.

In both cases, make sure the character is using the Chatterbox FAL model. Other voice models can treat these tags differently.

Good Tags to Start With

Use Chatterbox tags for moments where the nonverbal sound is part of the scene. These are reliable starting points:

• <sigh> for resignation, relief, exhaustion, or a quiet emotional turn.
• <chuckle> for restrained amusement or a line that should not become a full laugh.
• <laugh> when laughter is the point of the beat.
• <gasp> for surprise, fear, or sudden realization.
• <groan> for pain, annoyance, dread, or reluctant effort.

A Good Workflow

Do not tag everything. First generate a clean version of the scene. Listen for the lines where the human beat is missing or the reaction feels too flat. Add tags only to those lines, then regenerate the affected clips or the scene.

For a long scene, a handful of Chatterbox tags is usually enough. The model can still infer tone from punctuation and context, so reserve explicit tags for beats that need a clear sound.

Common Mistakes

• Using square brackets: use <sigh> for Chatterbox, not [sighing].
• Tagging a line that does not use Chatterbox FAL: other models may read the tag as literal text or ignore it.
• Stacking too many effects: one tag near the beat is usually clearer than chaining several tags into every sentence.
• Using tags instead of performable words: tags help, but the surrounding text, punctuation, and rhythm still shape the result.

Chatterbox Emotion Tag FAQ