Wiro - Realtime Conversational AI by ElevenLabs icon

Wiro - Realtime Conversational AI by ElevenLabs

Build real-time conversational AI with ElevenLabs' voice agent. Configure voice, language, and

Wiro - Realtime Conversational AI by ElevenLabs

AI-generated

Overview

This node integrates with ElevenLabs' Wiro API to create a real-time conversational AI voice agent. It allows users to configure various parameters such as the AI agent's voice, language, behavior instructions, text-to-speech model, and session settings. The node initiates a conversational AI session, waits for the session to complete by polling the task status, and outputs the session result including a task ID, status, and a URL to the conversation recording or result. This node is useful for building interactive voice-based AI applications, such as virtual assistants, customer support bots, or interactive voice response systems.

Use Case Examples

  1. A customer support bot that uses a friendly British voice to interact with users in real-time.
  2. An interactive storytelling application where the AI narrates stories with a deep narrator voice.
  3. A multilingual conversational agent that switches languages based on user input and uses a high-quality text-to-speech model.

Properties

Name Meaning
Voice The voice of the AI agent, selectable from predefined voice options such as British, American, Australian, etc.
System Instructions Instructions that define the AI agent's behavior and personality.
First Message The greeting message the agent says when the conversation starts.
Language Language for the AI agent, with options including Arabic, English, French, German, and more.
TTS Model Text-to-speech model used for voice synthesis, options include Flash V2.5 (fastest), Multilingual V2 (highest quality), and Turbo V2 (recommended for low latency).
Turn Timeout (Seconds) How long the agent waits for the user to respond before re-prompting, default is 7 seconds.
Silence Auto Close (Seconds) Auto-close session after this many seconds of user silence, default is 30 seconds.
Max Session Duration (Seconds) Maximum total session length in seconds, default is 600 seconds (10 minutes).
Turn Eagerness How eagerly the agent jumps in to respond; options are High (fast-paced), Low (patient), and Normal.
Auto End Call Allows the agent to automatically end the call when the conversation is complete; can be enabled or disabled.
Speech Speed Speech speed multiplier; 1 is normal speed, lower is slower, higher is faster.
Voice Stability Voice consistency; higher values mean more stable but monotone voice, lower values mean more expressive but variable voice.
Voice Similarity How closely the output matches the selected voice; higher values mean more similarity.
Streaming Latency Optimization Trade-off between audio quality and response speed; higher values mean faster response but lower quality.
Output Audio Format Audio format for AI voice output; must match the frontend audio player sample rate.
Input Audio Format Audio format for user microphone input; must match the frontend audio recorder sample rate.
Server Tools (JSON) Optional JSON array of webhook tools the agent can call during conversation, allowing integration with external APIs or services.

Output

JSON

  • taskid - The unique identifier of the conversational AI task.
  • url - URL to the conversation recording or result.
  • status - Status of the task, e.g., 'completed' or 'failed'.

Dependencies

  • ElevenLabs Wiro API

Troubleshooting

  • If the node throws an error about missing task ID or socket access token, check API credentials and ensure the Wiro API is reachable.
  • If the conversation does not complete, verify network connectivity and API limits.
  • Ensure the audio format settings match the frontend audio player and recorder sample rates to avoid audio issues.

Links

Discussion