# Deepgram MCP for AI Agents AI Agent Connect > Deepgram MCP turns speech into text and text into speech. It lets your AI agent handle high-speed transcription from URLs and generate natural-sounding voiceovers using neural networks. You get a direct line to Deepgram's latest models for production-grade audio processing without the manual overhead. ## Overview - **Category:** ai-frontier - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_b4nB7qQlpNhKdRs2pTzPWFY3jI1pKu8WFxapjmwu/ai-agent-connect - **Tags:** speech-to-text, text-to-speech, transcription, diarization, audio-processing, neural-networks ## Description You can turn raw audio into structured text or create high-quality voiceovers without leaving your current workspace. Instead of downloading files and uploading them to a web portal, you just give your agent a link. It handles the transcription, manages speaker labels, and returns the text ready for use. If you're building a voice-enabled app, you can use this same connection to turn your written content into human-like speech. It takes the headache out of choosing the right models or watching your usage limits. You can manage your project keys and monitor your API consumption through simple conversation. This makes it a core piece of the stack when you're building media tools or scaling research. You'll find this connection is especially useful when you have it organized in the Vinkius catalog with your other production tools. It's about moving from manual file management to a system where your agent acts as your dedicated audio engineer. ## Tools ### list_available_models See which high-performance models are ready for your project. This helps you pick the right balance of speed and accuracy. ### get_project_usage Check your current API limits and remaining quota. It keeps you on top of your costs and usage across different projects. ### list_api_keys View all your active API keys in one place. It makes it easy to manage credentials for different environments. ### list_deepgram_projects Get a full list of your active projects. Use this to see what's currently running in your Deepgram account. ### convert_text_to_speech Turn written text into natural human voices. This tool generates the audio files you need for voiceovers or apps. ### transcribe_audio_url Send a link to an audio file and get a text transcript back. It handles the processing so you don't have to upload anything. ## Prompt Examples **Prompt:** ``` I have a recording of a podcast at this link: https://example.com/podcast_01.wav. Can you get the transcript for me? ``` **Response:** ``` I've started the transcription for that podcast link. **Processing Details:** * **Model:** Nova-3 (High-fidelity) * **Status:** Converting audio to text... I'll provide the full transcript with speaker diarization as soon as it's ready. Would you like me to summarize the key points once I have it? ``` **Prompt:** ``` Turn this into a voiceover: 'Welcome to our new series on neural networks. We explore how machines learn to think.' ``` **Response:** ``` I've generated the voiceover for you using the high-speed Aura engine. **Audio Details:** * **Content:** "Welcome to our new series..." * **Format:** Natural human speech * **Status:** Ready for playback You can now access the audio file. Do you need me to generate any other segments? ``` **Prompt:** ``` How much of our Deepgram quota have we used this month? ``` **Response:** ``` Here is the current usage for your projects: | Project Name | Minutes Used | Request Count | | :--- | :--- | :--- | | Production_App | 450 mins | 1,200 | | Research_Batch | 120 mins | 450 | **Status:** You have plenty of room left for this week's tasks. ``` ## Capabilities ### Convert audio links to text Give your agent a URL and get a high-fidelity transcript with speaker identification. ### Generate human-like voiceovers Turn any block of text into natural-sounding audio files using high-speed neural engines. ### Browse available speech models See every model option to find the best balance of speed and accuracy for your specific project. ### Monitor API usage and limits Check your current minute consumption and request counts across multiple projects instantly. ### Manage project credentials Retrieve and organize your active API keys directly through your agent to keep things secure. ## Use Cases ### Automating video subtitles A content creator gives their agent a YouTube link and asks for a transcript. The agent uses transcribe_audio_url to get the text and speaker labels. ### Building a voice-enabled app A developer asks their agent to create a voiceover for a new feature. The agent uses convert_text_to_speech to generate the audio file. ### Scaling interview research A researcher provides a folder of links to recorded interviews. The agent processes them all to create a searchable text database. ### Managing production costs A project lead asks their agent to check the usage for the last week. The agent uses get_project_usage to report on the spending. ## Benefits - Faster transcription with transcribe_audio_url means you get text from long recordings in seconds instead of minutes. - Natural speech synthesis via convert_text_to_speech lets you create high-quality voiceovers that don't sound like robots. - Easy model selection with list_available_models ensures you always use the best tech for your specific audio needs. - Real-time monitoring with get_project_usage prevents you from hitting limits unexpectedly by showing your current spend. - Better security and organization with list_api_keys lets you manage your credentials without jumping between browser tabs. - Simplified project management with list_deepgram_projects gives you a birds-eye view of all your active audio workflows. ## How It Works The bottom line is you get a direct way to process audio and speech through your AI agent. 1. Grab your API key from the Deepgram Console. 2. Connect this Connector to your AI client like Claude or Cursor. 3. Ask your agent to transcribe a link or create a voiceover from text. ## Frequently Asked Questions **Can I use the Deepgram MCP to transcribe long audio files?** Yes, you can send links to long audio or video files. Your agent will handle the transcription and return the full text, including speaker labels, without you needing to upload anything manually. **Does the Deepgram MCP support different types of voices?** Yes, it gives your agent access to various neural speech models. You can choose different styles and tones to create high-quality voiceovers for your specific projects. **How does the Deepgram MCP help with my API costs?** It allows your agent to check your current usage and limits in real-time. This helps you monitor your spending and stay on top of your quota across all your different projects. **Is the Deepgram MCP good for real-time transcription?** It's designed for high-speed processing using the latest models. It's great for quickly turning recorded audio into usable text for research, content creation, or app development. **Can I manage my Deepgram keys through the Deepgram MCP?** Yes, you can use your agent to list and organize your active API keys. This makes it much easier to manage different credentials for your various development environments. **What happens if I give the Deepgram MCP a broken URL?** Your agent will attempt to access the link and let you know if there's an issue. It handles the communication with the audio service so you don't have to troubleshoot errors manually. **How do I get a Deepgram API Key?** Log in to the Deepgram Console, navigate to the **API Keys** section, and create a new key with the necessary permissions. **What is the Nova-3 model?** Nova-3 is Deepgram's latest state-of-the-art transcription model, offering unmatched speed and accuracy for real-world audio. **Can I synthesize speech in different voices?** Yes! The `convert_text_to_speech` tool allows you to specify models like `aura-asteria-en` or `aura-orion-en` for different vocal profiles.