Eleven v3 multi-person dialogue
Select a library voice for every turn, direct the performance, and generate one naturally connected scene.
Click a documented cue to insert it into the active line. Custom natural-language bracket directions are also supported.
… or [pause] for space, CAPITALS for emphasis, and punctuation or line breaks for rhythm. Combine cues such as [British accent] [exasperated]. Eleven v3 does not use SSML.ElevenLabs Music
Generate an original song from a detailed prompt. Avoid artist names, copyrighted lyrics, or requests to imitate a specific recording.
ElevenLabs Sound Effects
Describe a one-shot, ambience, Foley sequence, loop, braam, whoosh, drone, glitch, or musical element.
1. Upload a voice sample
10 seconds – 5 minutes. mp3, m4a, or wav. Max 20 MB.
2. Speak with the cloned voice
Cloned voices expire after 7 days if unused.