# Inworld AI MCP for AI Agents AI Agent Connect > Inworld AI MCP lets your agents speak with lifelike voices, clone specific audio samples, and manage complex character behaviors. It connects your AI client to Inworld's high-fidelity voice synthesis and character orchestration tools so you can build NPCs that sound human, not like robots. ## Overview - **Category:** artificial-intelligence - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_3lkEwqqakpRf4XlQXHb3G8jJBe7cyrieiM1PCdg3/ai-agent-connect - **Tags:** text-to-speech, voice-cloning, ai-characters, conversational-ai, speech-synthesis ## Description You want your characters to sound like they actually have a personality, not like a generic text-to-speech engine. This Connector brings Inworld's voice synthesis and character orchestration directly into your agent's workflow. Instead of just outputting text, your agent can now generate high-quality speech in real-time or create unique personas by describing them in plain English. You can take a short audio clip of a person speaking and turn it into a usable voice for your project, or build out complex logic to handle how different characters respond to the same situation. It handles the heavy lifting of managing voice assets and routing conversations so you can focus on the actual storytelling. When you find this Connector in the Vinkius catalog, you're getting a way to make your agents feel less like scripts and more like living entities. It's about moving past the robotic wall and into something that feels genuine. You get to focus on the narrative while the underlying tech handles the nuances of tone, pitch, and character logic. ## Tools ### delete_router Remove an existing LLM router from your workspace. Use this to clean up old character logic. ### delete_voice Delete a voice asset you no longer need. This keeps your workspace organized. ### get_router View the specific details of an LLM router. Use this to check your current routing configurations. ### get_voice Look up the properties of a specific voice asset. This helps you find the right ID for your project. ### list_models See which LLM models are available for use. This helps you pick the best brain for your character. ### list_routers See a list of all routers in your workspace. Use this to keep track of your active character logic. ### list_tts_voices View a list of older TTS voices. This is for legacy voice support in your workspace. ### list_voices See all the voices currently in your workspace. Use this to browse your available audio assets. ### publish_voice Move a draft voice into your published library. This makes the voice available for your agent to use. ### synthesize_speech_stream Generate speech as a stream for faster delivery. This is the best choice for real-time interaction. ### synthesize_speech_sync Generate a full audio file of speech at once. Use this for pre-rendered dialogue or long narrations. ### transcribe_audio Turn an audio file into a text string. This lets your agent understand spoken input from a user. ### update_router Edit the settings of an existing LLM router. Use this to tweak character behavior on the fly. ### update_voice Modify the details of a voice you already created. This is handy for refining a persona's tone. ### chat_completions Get responses from an LLM via a character router. This is the primary way to get intelligent, routed responses. ### clone_voice Create a custom voice by uploading audio samples. This is great for replicating a specific person's tone. ### create_realtime_call Set up a WebRTC call for live voice interaction. Use this for low-latency, two-way conversations. ### create_router Build an LLM router to manage complex character logic. It helps your agent decide how to respond based on context. ### design_voice Create a new voice by describing its characteristics in text. Just tell the agent what the character sounds like. ## Prompt Examples **Prompt:** ``` Make a voice for a tiny, energetic forest sprite. ``` **Response:** ``` I've designed a new voice for a 'tiny, energetic forest sprite' using `design_voice`. It has a high-pitched, fast-paced tone. Would you like me to publish it to your workspace? ``` **Prompt:** ``` Can you turn this audio clip into a voice for my main character? ``` **Response:** ``` I've started the process with `clone_voice`. Once the audio sample is processed, you'll have a high-fidelity clone of that character's voice ready for use. ``` **Prompt:** ``` How do I make the NPC sound more annoyed? ``` **Response:** ``` I can update the character's personality and voice settings. I'll use `update_router` to adjust the logic and `update_voice` to tweak the emotional delivery. ``` ## Capabilities ### Clone specific voices Turn audio samples into unique voice assets for your characters. ### Design voices with text Create new personas by describing their traits in plain English. ### Stream high-quality speech Get low-latency audio output for real-time conversations. ### Route character logic Manage how different characters respond to the same input. ### Transcribe spoken audio Convert audio files into text for your agent to process. ### Manage voice libraries Organize and publish your workspace's voice assets. ## Use Cases ### Dynamic NPC Dialogue A game dev uses create_router to make an NPC choose between being helpful or hostile based on a player's previous actions. ### Automated Voiceovers A content creator uses design_voice to make a grumpy wizard and then synthesize_speech_sync to generate a 10-minute story. ### Real-time Calls A developer uses create_realtime_call to let users talk to a customer service agent with almost no delay. ### Audio Analysis An engineer uses transcribe_audio to let an agent listen to a user's uploaded voice memo and summarize the main points. ## Benefits - You can create unique personas by using design_voice to describe exactly how a character should sound. - Real-time interactions feel smoother because synthesize_speech_stream allows for low-latency audio output. - Managing complex characters is easier with create_router, which handles the logic of who speaks and when. - You can quickly turn audio samples into usable assets with clone_voice for consistent character voices. - Transcribe_audio lets your agent understand spoken input, making your system truly multimodal. - You can organize your entire library of assets using list_voices and publish_voice to keep your project clean. ## How It Works The bottom line is you get a direct bridge between your agent's brain and Inworld's high-fidelity voice engine. 1. Subscribe to the Inworld AI MCP and grab your API keys from the Inworld Studio. 2. Plug those credentials into your AI client's connection settings. 3. Ask your agent to create voices, route logic, or generate speech directly. ## Frequently Asked Questions **Can I use Inworld AI MCP to make my characters sound like real people?** Yes. It uses high-fidelity synthesis and voice cloning to move past the robotic feel of standard TTS. **How does Inworld AI MCP handle complex character personalities?** It uses a routing system that allows your agent to decide how a character should behave based on context. **Can I clone my own voice using Inworld AI MCP?** You can upload audio samples to clone_voice to create a custom persona for your project. **Does Inworld AI MCP support real-time speech?** It supports streaming synthesis, which is perfect for low-latency conversations. **Can I use Inworld AI MCP for game development?** It's a primary choice for devs who want dynamic, voiced NPCs that respond to players. **What happens if I want to change a character's voice later?** You can use update_voice to modify existing assets without having to start over. **Can Inworld AI MCP transcribe audio files?** Yes, it includes a tool to turn spoken audio into text for your agent to process. **How can I create a custom voice using only a text description?** You can use the `design_voice` tool. Simply provide a prompt like 'Warm, friendly male voice' and a preview text. The tool will generate voice options that you can later publish to your library. **What is the difference between synchronous and streaming speech synthesis?** Use `synthesize_speech_sync` to receive the full audio file once processing is complete. Use `synthesize_speech_stream` for real-time applications where you want to receive audio chunks as they are generated for lower latency. **Can I manage multiple AI characters or models through this server?** Yes. You can use `list_routers` and `get_router` to manage your orchestration layers, and `list_models` to see available AI models in your Inworld workspace.