Tavus

Docs

• Authentication: Generate an API key in the Developer Portal and send it on each request in the x-api-key header.
• Create conversation: This endpoint starts a real-time video conversation with your AI replica, powered by a persona that allows it to see, hear, and respond like a human.
• Delete Conversation: This endpoint deletes a single conversation by its unique identifier. Use this for destructive data removal. For normal call cleanup when a user leaves or a session is finished, use End Conversation instead.
• End Conversation: This endpoint ends a single conversation by its unique identifier. Use this for routine call cleanup when a user leaves or your app no longer needs the room. To destructively remove conversation data, use Delete Conversation.
• Get Conversation: This endpoint returns a single conversation by its unique identifier.
• List Conversations: This endpoint returns a list of all Conversations created by the account associated with the API Key in use.
• Create Document: Upload documents to your knowledge base for personas to reference during conversations.
• Delete Document: Delete a document and its associated data using its unique identifier.
• Get Document: Retrieve detailed information about a specific document using its unique identifier.
• List Documents: Retrieve a list of documents.
• Update Document: Update a document's document_name and tags.
• Recrawl Document: Trigger a recrawl of a website document to fetch fresh content.
• Create Guardrails: Create a new guardrail. Guardrails provide strict behavioral boundaries that are enforced throughout a conversation.
• Delete Guardrails: Delete a single guardrail by its unique identifier. Personas with this guardrail attached via guardrail_ids will have the reference removed automatically.
• Get Guardrails: Retrieve a single guardrail by its unique identifier.
• List Guardrails: Return a flat list of guardrails owned by the caller. Pass legacy=false — recommended for all new integrations.
• Patch Guardrails: Update specific fields of a guardrail using JSON Patch operations. Paths must match the current document shape — compare against the response from Get Guardrails before patching.
• Create Objectives: This endpoint creates objectives for a persona.
• Delete Objective: This endpoint deletes a single objective by its unique identifier.
• Get Objective: This endpoint returns a single objective by its unique identifier.
• Get Objectives: This endpoint returns a list of all objectives.
• Patch Objective: This endpoint allows you to update specific fields of an objective using JSON Patch operations.
• Overview: Discover the Tavus API — build a real-time, human-like multimodal video conversation with a replica.
• Create Persona: Creates a persona that configures how a replica behaves and sounds in CVI for every conversation that uses that persona.
• Delete Persona: This endpoint deletes a single persona by its unique identifier.
• Get Persona: This endpoint returns a single persona by its unique identifier.
• List Personas: This endpoint returns a list of all Personas created by the account associated with the API Key in use.
• Patch Persona: This endpoint allows you to update specific fields of a persona using JSON Patch operations.
• Create Replica: Creates a new replica from a training video or image URL for use in conversations—see Which training path? for preparation, consent, and media requirements.
• Delete Replica: Deletes a Replica by its unique ID; deleted replicas cannot be used in a conversation.
• Get Replica: This endpoint returns a single Replica by its unique identifier.
• List Replicas: This endpoint returns a list of all Replicas created by the account associated with the API Key in use.
• Rename Replica: This endpoint renames a single Replica by its unique identifier.
• Create Pronunciation Dictionary: Create a pronunciation dictionary with custom rules for controlling how words are spoken. Rules are automatically synced to both Cartesia and ElevenLabs so they work regardless of which TTS engine your persona uses.
• Delete Pronunciation Dictionary: Permanently delete a pronunciation dictionary and remove it from all linked personas.
• Get Pronunciation Dictionary: Retrieve a pronunciation dictionary by its ID, including all rules.
• List Pronunciation Dictionaries: List all pronunciation dictionaries for the authenticated user with pagination.
• Update Pronunciation Dictionary: Update a pronunciation dictionary's name or rules using JSON Patch format (RFC 6902). Supported mutable fields are name and rules.
• Generate Video: This endpoint generates a new video using a Replica and either a script or an audio file.
• Delete Video: This endpoint deletes a single video by its unique identifier.
• Get Video: This endpoint returns a single video by its unique identifier.
• List Videos: This endpoint returns a list of all Videos created by the account associated with the API Key in use.
• Rename Video: This endpoint renames a single video by its unique identifier.
• List Voices: Returns available stock voice_name values and their linked replica metadata. When you Create Replica with train_image_url (image-to-replica), voice_name is required—use this list to pick a valid slug and to preview options.
• Agents & automation: How to access machine-readable docs and API artifacts for developers, IDEs, and automation - llms.txt, OpenAPI, Agent Skills, and MCP.
• Changelog
• Blocks: High-level component compositions that combine multiple UI elements into complete interface layouts
• Components: Learn about our pre-built React components to accelerate integrating the Tavus Conversational Video Interface (CVI) into your application.
• Hooks: See what hooks Tavus supports for managing video calls, media controls, participant management, and conversation events.
• Overview: Learn how our Tavus Conversational Video Interface (CVI) Component Library can help you go live in minutes.
• Server: Server-side helpers for managing Tavus conversation lifecycle without leaking your API key to the browser.
• Audio-Only Conversation: Start a conversation in audio-only mode, perfect for voice-only or low-bandwidth environments.
• Background Customizations: Apply a green screen or custom background for a personalized visual experience.
• Call Duration and Timeout: Configure call duration and timeout behavior to manage how and when a conversation ends.
• Closed Captions: Enable closed captions for accessibility or live transcription during conversations.
• Participant Limits: Control the maximum number of participants allowed in a conversation.
• Private Rooms: Create authenticated conversations with meeting tokens for enhanced security.
• Overview: Learn how to customize identity and advanced settings for a conversation to suit your needs.
• Customer Support: Engage in real-time customer support and sales conversations with the Customer Support persona (Gloria).
• Interviewer: Engage with the Interviewer persona to run structured, conversational screening interviews.
• Sales Coach: Engage with the Sales Coach persona to simulate real-time sales conversations.
• Sales Development Rep: Engage with Anna, the Tavus sales development rep persona.
• FAQs: Frequently asked questions about Tavus's Conversational Video Interface.
• Guardrails: Guardrails provide your persona with strict behavioral guidelines that will be rigorously followed throughout every conversation.
• Interaction Events: Control CVI conversations by sending and listening to interaction events.
• Knowledge Base: Upload documents to your knowledge base for personas to reference during conversations.
• Language Support: Customize the conversation language using full language names supported by Tavus TTS engines.
• Memories: Memories let personas remember information across conversations, allowing participants to have personalized, flowing conversations across multiple sessions.
• Mobile: Ship Tavus CVI on iOS, Android, React Native, and mobile web using Daily's mobile SDKs and the same conversation URLs as web.
• What is CVI?: CVI enables real-time, human-like video interactions through configurable lifelike replicas.
• Conversational Flow: Learn how to configure the Conversational Flow layer to fine-tune turn-taking and interruption handling behavior.
• Large Language Model (LLM): Learn how to use Tavus-optimized LLMs or integrate your own custom LLM.
• Tool Calling for LLM: Set up tool calling to trigger functions from user speech using Tavus-hosted or custom LLMs.
• Objectives: Objectives are goal-oriented instructions to define the desired outcomes and flow of your conversations.
• Overview: Define how your persona behaves, responds, and speaks by configuring layers and pipeline modes.
• Perception: Learn how to configure the perception layer with Raven to enable real-time visual and audio understanding.
• Tool Calling for Perception: Configure tool calling with Raven to trigger functions from visual or audio input.
• Pronunciation dictionaries: Control how your persona pronounces specific words, names, and terms during conversations.
• Stock Personas: Tavus offers pre-built personas to help you get started quickly.
• Speech-to-Text (STT): Configure the STT layer to select an STT model, improve transcription accuracy, and optimize for your target languages.
• Text-to-Speech (TTS): Discover how to integrate custom voices from third-party TTS engines for multilingual or localized speech output.
• CVI App: AI Prompt: Copy-paste checklist for Cursor, Copilot, or other AI coding agents to scaffold React (TypeScript, Vite) with Tavus CVI and @tavus/cvi-ui.
• CVI App Quickstart: Create a server-authenticated Tavus CVI conversation and embed it in a web app with the returned conversation_url.
• Conversation Recordings: Store conversation recordings in your own S3, GCS, or Azure Blob storage. Federated identity — no secrets shared with Tavus.
• Customize Conversation UI: Experience a conversation in a custom Daily UI — styled to match your preference.
• API Conversation Quickstart: Create your first persona using the full pipeline and start a conversation in seconds.
• Emotion Control with Phoenix-4: Unlock emotionally expressive facial movements and micro-expressions using Phoenix-4 replicas.
• Pipeline Modes: Run CVI with the full Tavus pipeline, Echo mode, or integrate via LiveKit and Pipecat.
• Errors and Status Details: Identify errors and status details encountered when using the Tavus platform.
• Append Conversational Context Interaction
• Echo Interaction
• Interrupt Interaction
• Overwrite Conversational Context Interaction
• Perception Analysis Event
• Perception Tool Call Event
• Text Respond Interaction
• Sensitivity Interaction
• Started/Stopped Speaking Event
• Tool Call Event
• Utterance Event
• Utterance Streaming Event
• Example Projects
• Embed Conversational Video Interface: Learn how to embed Tavus's Conversational Video Interface (CVI) into your site or app.
• LiveKit Agent: Integrate a Tavus Replica into LiveKit as the conversational video avatar.
• Pipecat: Integrate a Tavus Replica into your Pipecat application as a participant or a video feed for the bot.
• Introduction: Leverage Tavus tools and guides to give your AI Agent real-time human-like perception and presence, bringing the human layer to AI.
• Models
• Reducing Join Latency: Strategies to reduce perceived and actual latency when starting Tavus conversations
• Objectives Prompting Guide: Write clear, testable prompts for objectives with practical examples.
• Persona Strategies: Two approaches to using personas at scale: reuse with conversational context vs. create-per-session. Choose based on whether you need different data or different behavior per conversation.
• Prompting Guide: The recommended structure for writing system prompts for Tavus CVI personas - identity, style, behaviors, and guardrails that work in real-time video conversations.
• Tavus Tool Calling: Key guidelines for implementing reliable tool calls in Tavus conversational agents
• What are PALs?: AI companion experiences with text, voice, and video—and how they relate to Tavus developer products.
• Image to Replica: Create a Phoenix replica from a headshot image and stock voice_name using the Create Replica and Get Replica APIs.
• Overview: Understand Tavus replicas, learn how to create your own using a video or an image, or explore ready-to-use stock replicas.
• Replica FAQs
• Stock Replicas: Start a conversation instantly with our ready-to-use stock replicas - no training required. Stock replicas are prebuilt, high-quality AI agents recorded with real actors and designed for natural, expressive communication. They enable lifelike interactions with dynamic facial expressions, smooth turn…
• Training from a Video: Phoenix-4 video training requirements, recording structure, and quality guidelines.
• Training from an Image: Upload a headshot and select a voice_name to create a replica without recording a training video.
• Video to Replica: Create a Phoenix replica from a training video using the Create Replica and Get Replica APIs.
• Which Training Path?: Compare video-based and image-based Phoenix replica training and open the guide that matches your workflow.
• Troubleshooting: Find solutions to common problems and get back on track quickly with our troubleshooting guides.
• Background Customizations: Customize AI video backgrounds with transparency, scrolling websites, or custom video sources.
• Overview: Generate high-quality async videos from a script and a replica—separate from real-time CVI.
• Quickstart: Async Video API walkthrough—render a file from script or audio with a replica; not real-time CVI.
• Webhooks and Callbacks: Set up a webhook server to generate a callback URL that receives event notifications from Tavus API.

OpenAPI Specs

• openapi