"The concept of thinking or introspection" matching MCP servers:

MCP server for audio transcription using OpenRouter models, supporting verbatim, cleaned, and custom transcription modes.

Enables AI assistants to download YouTube videos and transcribe them locally using OpenAI's Whisper speech recognition model.

A self-hosted MCP server that connects AI clients to local media-generation backends like ComfyUI and Blender, and provides built-in utilities for video/audio analysis, editing, subtitles, speech, and scene detection via a persistent file cache.

Enables hair colourists to capture a client's colour formula by voice, retrieve it on the next visit, and scale the mix to today's bowl, with per-client memory and clarification when names are ambiguous.

Enables AI clients to transcribe YouTube links and uploaded audio or video files, then retrieve the resulting transcripts as text, SRT, VTT, or JSON with timestamps and optional speaker labels. Also supports paging through long transcripts, checking upload progress, listing or deleting jobs, and viewing remaining account minutes.

Enables MCP clients such as Claude Code, Codex and Cursor to retrieve timestamped transcripts of public YouTube videos, either from cache, from native captions, or via asynchronous audio-generation jobs that are polled by job ID. It returns the source URL, selected language and millisecond-offset segments without summarizing the video itself.

MCP server for SpeechWeave transcription, enabling AI assistants to transcribe local files and URLs via wait-first or async tools.

Voice input for Claude Code — speak Vietnamese or English; hold a hotkey, speak, release, and the transcribed text is cleaned and pasted into the input box for editing before sending.

MCP server for Whipscribe — transcribe audio and video from a URL or local file via Claude Desktop, Claude Code, Cursor, Windsurf, or any MCP-compatible client.

Enables an assistant to transcribe video or podcast recordings, then read, search, summarize, and manage the resulting transcripts through MCP tools.

Enables agents to submit audio files to the local Meeting Transcriber macOS app, retrieve diarized transcripts, assign speaker names, and start or stop automatic meeting detection over its localhost automation API. Everything runs on-device, so no audio or transcripts leave the Mac.

Bridges MCP-capable AI tools such as Claude Code, Cursor, and Codex CLI to the Prodius desktop app, letting an agent pop a question on screen and capture your spoken or typed reply without switching back to the terminal. It also sends fire-and-forget desktop notifications and can answer permission prompts by voice.

Enables an AI assistant to place user-approved phone calls in the callee's language through your own Bland or Twilio account, with language detection, script generation, cost estimates, and safety guard rails. Includes a preview-then-confirm gate, plus call status, transcripts, and summaries.

Enables AI assistants to generate animated overlays and edit short videos — cutting clips together, removing pauses, adding captions, and exporting overlays as MP4, MOV or WebM with transparent or green-screen backgrounds. All rendering, transcription and media handling happen locally on the user's own computer.

Enables local, on-device transcription of local files or URLs, returning a job ID and making transcripts, segments, and full-text search available as processing completes. Uses Whisper large-v3 with SQLite storage and stdio transport, so audio and video never leave the machine.

Enables MCP clients such as Grok Bot to create and join GifOS meetings, send in a visible note-taking bot that transcribes participants locally via Whisper, post chat content, and read back or save the transcript. Everything, including the browser, bot credentials, and transcripts, runs on the user's own machine with no GifOS account or API key.

Enables Claude to fetch the spoken text of YouTube videos from any URL or video ID, listing available captions and languages and paging through long transcripts. It also handles proxy configuration to work around YouTube IP blocks.

Turns Claude or ChatGPT into a live system design interviewer that watches a hand-drawn whiteboard, reads the diagram as structured data or an image, hears spoken answers, pins questions on components, and grades the session. Everything runs locally, so users can practice mock interviews on their own machine without API keys.

Enables local, upload-free analysis of publicly viewable videos through tools that pull metadata, chapter plans, platform subtitles or speech-to-text transcripts, sampled frames and contact sheets, 527-class sound event tags, and comment analysis. It downloads only the minimum necessary bytes, so users can inspect long videos section by section, job-poll lengthy runs, and check the local model/environment setup without sending media to any cloud service.

Extracts transcripts or spoken captions from multi-platform video links (Bilibili, YouTube, Douyin) and returns concise Markdown summaries or full transcripts, saving tokens by keeping full text out of the conversation.

Enables AI agents on macOS to securely read and search the local Messages database, catch up on missed messages via a persistent inbox, and send texts or files to allowlisted chats, with optional voice note transcription and text-to-speech.

Enables Claude to capture both sides of macOS calls and meetings from apps like WhatsApp, Teams, Zoom, and browsers as aligned microphone and system-audio tracks, then transcribe them locally with Whisper. Supports speaker separation, custom vocabulary, and importing existing audio or video files, with everything processed on-device so no audio leaves the Mac.

Enables transcription of YouTube videos by using available captions or falling back to actual audio with local Whisper ASR and multiple free providers. It supports job status, transcript retrieval and search, audio excerpts, and export formats while keeping source links and timestamps.

Provides intelligent transcript processing capabilities for Claude, featuring natural formatting, contextual repair, and smart summarization powered by Deep Thinking LLMs.

Lets Grok Bot or any MCP client schedule a GifOS broadcast with complete invite details (link, start and end times, visit link, ready-to-send text), run the host's booth to go live, play a prerecorded video and end the stream, and seat a note-taking bot that transcribes what is said on stage. It runs entirely on the user's own machine, broadcasting over GifOS's peer-to-peer fan-out with no account, API key, or hosted endpoint.

A MCP server that enables transcription of audio files using OpenAI's Speech-to-Text API, with support for multiple languages and file saving options.

MCP server for a Polish language tutor that provides interactive exercises, concept explanations, pronunciation practice, and progress tracking using a local LLM and a knowledge graph.

Best VoiceAI White Label Platform for Agencies with Full MCP Connectivity to Any Platform for Custom VoiceAi & AI Chat Integrations and You Keep 100% of the Profits

Enables full coverage of the ElevenLabs API through 390 auto-generated tools, handling file uploads for speech-to-text and voice cloning plus raw audio downloads. Supports domain filtering, dry-run guarding of credit-spending calls, and configurable output paths for generated media.

Know what your cloud agents did without reading logs. Heard MCP lets Claude, ChatGPT, Devin, Copilot and other cloud agents brief you out loud on your Mac, by name: what they got done, what's blocking them, or the decision they need. "Scout finished: the CI fix is merged." Answer out loud and the agent picks up your reply and keeps going. Hosted, OAuth sign-in, nothing to install.