# AssemblyAI MCP for AI Agents AI Agent Connect > AssemblyAI lets you turn audio files and video links into structured text. It handles the heavy lifting of speech-to-text, allowing your AI agent to read transcripts, check for accuracy with confidence scores, and organize your entire library of spoken content without you having to manually upload or manage files one by one. ## Overview - **Category:** ai-frontier - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_kHfyjES8DV8n2LXmFad8bSeOK9tKPV57hzP83TCE/ai-agent-connect - **Tags:** speech-to-text, transcription, audio-processing, natural-language-processing, ai-models ## Description Imagine you have a mountain of podcast recordings, meeting notes, and raw video files. Right now, you probably have to download them, upload them to a transcription service, wait for a notification, and then spend an hour cleaning up the text. It's a massive time sink. This Connector changes that. Instead of jumping between tabs, you just tell your AI client to get the transcript for that podcast link or summarize the last three meetings. The agent handles the connection to AssemblyAI, starts the job, and keeps an eye on the progress. When it's done, your agent can actually read the content for you, breaking it down into paragraphs or sentences so you can find exactly what you need. You can even ask it to double-check the accuracy of a specific section by looking at the confidence scores. It turns a manual multi-step process into a simple conversation. It's one of the best ways to manage audio data in the Vinkius catalog because it gives your agent real-time eyes on your spoken content. You stop being a file manager and start being a content strategist. ## Tools ### delete_transcript Removes a specific transcription record from your history. Use this to keep your workspace organized. ### get_transcript_paragraphs Retrieves the transcript formatted into paragraphs. It's perfect for reading long-form content. ### get_transcript_sentences Breaks the transcript down into individual sentences. Use this for precise text extraction or editing. ### get_transcript Fetches the full result of a completed transcription job. This is your primary way to see the final text. ### list_transcripts Shows a list of all your transcription jobs. Use it to track what's finished and what's still processing. ### transcribe_audio Starts a transcription job for a given audio or video URL. This is how you kick off the process. ## Prompt Examples **Prompt:** ``` Can you summarize the interview at https://example.com/interview.mp3? ``` **Response:** ``` I've started transcribing that interview for you. Once it's finished, I'll provide a summary of the key points and a list of the main topics discussed. **Status:** - **Job ID:** `as_998` - **Status:** Processing - **Estimated Time:** 2 minutes ``` **Prompt:** ``` What was the confidence level for the last transcript? ``` **Response:** ``` The transcript for the 'Marketing Sync' had an overall confidence score of **94%**. Most of it is very reliable, but there are a few segments in the middle with lower scores that might need a quick human check. ``` **Prompt:** ``` List my recent transcripts and tell me which ones are done. ``` **Response:** ``` **Recent Jobs:** | Job Name | Status | Confidence | | :--- | :--- | :--- | | Marketing Sync | Completed | 94% | | Product Demo | Processing | - | | Client Onboarding | Completed | 98% | | Weekly Standup | Completed | 96% | ``` ## Capabilities ### Start transcription from a URL Turn any audio or video link into a text transcript instantly. ### Get paragraph-level breakdowns Organize long transcripts into readable sections for easier navigation. ### Get sentence-level breakdowns Extract specific lines from a transcript for precise editing or quoting. ### Check transcription confidence Verify which parts of an audio file might need a human eye for accuracy. ### List all past jobs See your entire history of processed audio in one view. ### Delete old records Keep your transcription history clean by removing old jobs. ## Use Cases ### Podcast Show Notes A creator asks the agent to summarize the podcast at a specific link. The agent uses transcribe_audio and get_transcript_paragraphs to generate a bulleted summary. ### Meeting Audit An ops lead asks what was decided about the budget in last week's meeting. The agent finds the right job via list_transcripts and reads the text. ### Accuracy Check A researcher wants to know if a specific quote is correct. The agent pulls the transcript and highlights segments with low confidence scores. ### Bulk Transcription Management A developer asks the agent to list all transcripts from last month to audit their usage and find specific job IDs. ## Benefits - Stop manual uploads. Use transcribe_audio to start jobs directly from URLs, cutting out the need to download and re-upload files. - Get structured content. Use get_transcript_paragraphs to turn a wall of text into organized sections that your agent can actually navigate. - Verify accuracy quickly. Use get_transcript to see confidence scores, so you know exactly which parts of an audio file need a human double-check. - Manage history at scale. Use list_transcripts to keep track of every job you've ever run without leaving your chat interface. - Granular data extraction. Use get_transcript_sentences to pull out specific quotes or lines without having to scroll through a massive transcript. ## How It Works The bottom line is you get searchable, structured text from any audio link without touching a dashboard. 1. Add your AssemblyAI API key to your configuration. 2. Provide your AI client with a URL for an audio or video file. 3. Receive the processed text or a status update on the transcription progress. ## Frequently Asked Questions **How does AssemblyAI MCP handle my audio files?** It takes audio or video URLs and turns them into structured text transcripts. This means you don't have to manually upload files to a dashboard. **Can I use AssemblyAI MCP to summarize my podcasts?** Yes. Once the agent transcribes the podcast link, it can read the text and provide summaries, speaker labels, or specific highlights. **Does AssemblyAI MCP support video files?** Yes, it supports both audio and video URLs, making it useful for analyzing webinars, recordings, or social media clips. **How do I know if the transcription is accurate?** The Connector provides confidence scores for the text. Your agent can check these scores and alert you to specific parts that might need a human review. **Can AssemblyAI MCP organize my transcripts?** Yes, it can list your entire history of completed and active jobs so you can stay organized and find specific recordings quickly. **Does AssemblyAI MCP work with my existing files?** It works with any audio or video file hosted at a public URL. You just give your agent the link to start the process. **Can I delete old transcripts with AssemblyAI MCP?** Yes, you can ask your agent to remove specific transcription records to keep your history clean and manageable. **How do I find my AssemblyAI API Key?** Log in to your [**AssemblyAI dashboard**](https://www.assemblyai.com/app), and you will find your API Key on the main home page. Copy and paste it below. **What audio formats are supported?** AssemblyAI supports most common audio and video formats, including MP3, WAV, AAC, MP4, and others. Simply provide a public URL to the file. **Can the agent identify different speakers?** Yes. When starting a job via `transcribe_audio`, set the `speaker_labels` parameter to true. Your agent will return the text categorized by speaker ID.