TTS Eleven Labs
Posted: Sun Mar 08, 2026 6:29 pm
I just learnt something that for me anyway is pretty incredible. I have played with TTS and you will see some of this in Rise of Pax, but just learnt you can prompt Eleven Labs in incredibly powerful ways I didn't even now about, so hope this is useful to some here too -
Key Things You Can Do with Square Brackets in ElevenLabs
Emotions and delivery styles: [happy], [sad], [excited], [angry], [whispering], [shouting], [nervous], [thoughtful], [annoyed], [surprised], [dramatic], etc.
Non-verbal vocalizations: [laughing], [chuckles], [sighs], [groaning], [clears throat], [exhales sharply], [inhales deeply], [crying], etc.
Pauses and timing: [pause], [short pause], [long pause] (or similar descriptive ones).
Note: These are interpretive; for precise control, ElevenLabs often recommends <break time="1.5s" /> (SSML-style) instead, especially in older models or API use. But audio tags work well for natural-feeling breaks.
Overall direction or context: Things like [whisper], [football commentary style], or chained tags [whispering][pause] for combined effects.
Other tricks: Tags are not case-sensitive (e.g., [HAPPY] or [happy] both work), and you can chain them [laughing][excited]. Place them right before/after the text they affect for best results, like:
"[laughing] That's hilarious! [sighs] But seriously..."
Key Things You Can Do with Square Brackets in ElevenLabs
Emotions and delivery styles: [happy], [sad], [excited], [angry], [whispering], [shouting], [nervous], [thoughtful], [annoyed], [surprised], [dramatic], etc.
Non-verbal vocalizations: [laughing], [chuckles], [sighs], [groaning], [clears throat], [exhales sharply], [inhales deeply], [crying], etc.
Pauses and timing: [pause], [short pause], [long pause] (or similar descriptive ones).
Note: These are interpretive; for precise control, ElevenLabs often recommends <break time="1.5s" /> (SSML-style) instead, especially in older models or API use. But audio tags work well for natural-feeling breaks.
Overall direction or context: Things like [whisper], [football commentary style], or chained tags [whispering][pause] for combined effects.
Other tricks: Tags are not case-sensitive (e.g., [HAPPY] or [happy] both work), and you can chain them [laughing][excited]. Place them right before/after the text they affect for best results, like:
"[laughing] That's hilarious! [sighs] But seriously..."