---
title: "speech-elevenlabs"
description: "ElevenLabs speech provider: an agent's replies spoken by Eleven v4 Turbo (or v4, or Flash v2.5) with word-level caption timing, and live voice conversations through an ElevenLabs agent that hands repository work to your "
canonical: https://agentpluginsdirectory.com/plugins/speech-elevenlabs
last-updated: 2026-10-05
---

# speech-elevenlabs
ElevenLabs speech provider: an agent's replies spoken by Eleven v4 Turbo (or v4, or Flash v2.5) with word-level caption timing, and live voice conversations through an ElevenLabs agent that hands repository work to your coding agent. Choose it from a voice profile by naming the provider "elevenlabs".
- Slug: speech-elevenlabs
- Publisher: Dimension
- Repository: https://github.com/insodimension/dimension-community
- Manifest: packs/speech-elevenlabs/plugin.json
- Version: 0.2.1
- License: MIT
- Category (editorial): other
- Skills: 0
- MCP servers: 0
- Stars: 1
- Repository created: 2026-07-31
- Repository last pushed: 2026-10-05
- Publisher type: User
- Listing: https://agentpluginsdirectory.com/plugins/speech-elevenlabs
- Schema: https://agent-plugins.org/schemas/1.0.0/plugin.schema.json

## What speech-elevenlabs does, in the publisher's words

A speech provider for Dimension: an agent's replies spoken by ElevenLabs (Eleven v4 Turbo by default, or Eleven v4, or Flash v2.5) with word-level caption timing, and live voice conversations through an ElevenLabs agent that hands repository work to your coding agent. A voice profile picks it by naming the provider elevenlabs.

- The elevenlabs speech provider (src/index.ts), declared speak: true, converse: true, listen: false. Eleven v3 and v4 are served over the Text-to-Dialogue socket, which streams audio with character alignment a few hundred milliseconds after the first text; every other model is spoken one segment per HTTP request. The caller sees the same speaking session either way. Listening is not offered: pair it with an on-device listener.
- Live calls (src/converse.ts, src/convai.ts, src/agents.ts). The engine holds the ElevenLabs Agents socket and carries microphone audio up and the voice down.

From the project README, punctuation lightly normalized. Full text: https://raw.githubusercontent.com/insodimension/dimension-community/HEAD/packs/speech-elevenlabs/README.md