Skip to main content
This is a premium feature. Upgrade your plan to access it.
Audio tags are short directions in square brackets that change how the Eleven v3 voice model speaks. Use them to add emotion, whispering or shouting, laughter, pauses, accents, and character voices to text-to-speech. Audio tags work anywhere BrewBot reads text with the shared TTS voice:

Turn on Eleven v3

  1. Open Live Tools > TTS Settings.
  2. In the Voice card, set the voice model to Eleven v3.
  3. Add tags to chat messages, announcements, or story text.
With Eleven Multilingual v2, tags aren’t performed and may be read aloud or ignored.

How to write tags

Put a tag directly before the words it should affect. You can change direction several times in one message, even mid-sentence.
  • Tags use [square brackets].
  • Tags are written in plain English. There is no fixed list, so you can describe a delivery in your own words.
  • Combine directions in one tag with commas, such as [quietly, after a pause] or [angry, shouting].
  • Tags count toward message and announcement character limits.
  • A message needs spoken words besides its tags. A message made only of tags, such as // [laughs], is skipped.
Tags are prompts, not guaranteed effects. Results vary by voice and by the surrounding text, and not every tag works with every voice. Test a tag with your chosen voice before relying on it live.

Emotions

Change how the speaker feels.

Delivery and volume

Control how loud and how forcefully something is said. [whispers] [whispering] [quietly] [softly] [shouts] [shouting] [loudly] [screaming] [yelling] [understated] [emphasized] [stress on next word] [deliberate] [neutral tone] [deep voice] [robotic tone]

Laughter and reactions

Add laughs, breaths, and other non-verbal sounds.

Pacing and pauses

Shape timing for conversation or narration. [pause] [pauses] [long pause] [continues after a beat] [hesitates] [stammers] [rushed] [rapid-fire] [fast-paced] [slows down] [slowly] [drawn out] [repeats] [breathes] [deliberate]
Eleven v3 doesn’t support SSML break tags such as <break time="1s" />. Use [pause] or [long pause], or punctuation such as ellipses (...) and line breaks.

Accents

Change pronunciation and speaking style without translating the words. [American accent] [British accent] [Australian accent] [Irish accent] [Scottish accent] [Southern US accent] [New York accent] [Indian English] [French accent] [German accent] [Spanish accent] [Italian accent] [Russian accent] [strong French accent] [thick Irish accent] Use the pattern [strong X accent] or [thick X accent] to try other accents.

Character voices

Give the voice a persona or acting style. [pirate voice] [evil scientist voice] [childlike tone] [deep voice] [robotic tone] [fantasy narrator] [sci-fi AI voice] [classic film noir] [dramatic] [sarcastically] [matter-of-fact] [whiny]

Conversation and interruptions

Direct turn-taking when a message or story reads like a conversation. [interrupting] [overlapping] [cuts in] [interjecting] [jumping in] [starting to speak] [hesitates] [fast-paced] [flustered] [casual]

Sound effects

[applause] [clapping] [gunshot] [explosion] [fart] [sings] [woo] [swallows] [gulps]
Sound effect tags are experimental and vary a lot between voices. For a specific, repeatable sound, use Gift Sounds or Sticker Sounds instead.

Combined tags

Combine several directions in one tag for a more specific performance. These are starting points to experiment with, not presets.

Example script

This works well as a TTS Story or announcement:

Troubleshooting