ElevenLabsElevenLabs Dialogue
Multi-speaker dialogue from a single API call. Up to 10 speakers, per-speaker voice ids, and stitched conversational audio output.
ElevenLabs Dialogue is the multi-speaker endpoint. Up to 10 speakers, each with their own ElevenLabs voice id. The service stitches the conversation natively, so turn handoffs sound like a real chat instead of pasted-together TTS calls.
What it is good at
Scripted podcasts
Alternating hosts with distinct timbres in one API call.
Audiobook scenes
Multiple cast voices for dialogue passages.
Training simulators
Role-play lines with fixed casting per character.
Game barks
Short, turn-based dialogue between NPCs.
Product walkthroughs
Conversational tone instead of single narrator.
Language learning
Per-speaker stability for clarity in listening exercises.
Why Dialogue over stitched TTS calls
Head to head
Compared to stitching multiple TTS renders
- Native conversational pacing instead of mechanical turn boundaries.
- Single API call instead of N+1 calls plus stitching code.
- Per-speaker voice ids in one structured payload.
- Consistent room and ambience across speakers.
- Built-in stability control across the full conversation.
FAQs
People also ask
A dialogue endpoint that builds an array of {text, voice} turns. Each speaker uses an ElevenLabs voice id. The service renders one combined audio file for the conversation.
Fill Speaker 1 and Speaker 2 at minimum. Optional speakers 3-10 appear when you enable their toggles, each with voice and text fields.
No. Dialogue uses its own integration path. You set languages and stability in the dialogue form rather than the single-speaker TTS model list.
Use Dialogue when conversational pacing and voice handoff matter. Stitching separate TTS renders gives flat turn boundaries and inconsistent room. Dialogue handles the joins natively.
ElevenLabs Dialogue ships as a single API call with up to 10 distinct ElevenLabs voices. OpenAI's multi-speaker offerings sit inside their TTS pipeline. HeyGen is video-focused with a built-in voice list. Pick by ecosystem.