"Speech & Transcription" matching MCP servers:

Sub-150ms clean Web-to-Markdown extraction, autonomous agent live web search, and neural speech synthesis for Claude Desktop, Cursor, and Windsurf.

Provides browser automation, audio transcription, and LLM chat as MCP tools for any agent.

Records narrated walkthroughs of any webapp by orchestrating Playwright, speech-mcp TTS, and FFmpeg composition.

MCP server providing tools for web scraping, browser automation, computer vision, audio transcription, and RAG via API.

Enables users to look up a Douyin creator by name, public ID, or profile URL and receive the latest video link, like/comment/favorite/share counts, a full timestamped transcript with an AI-reviewed correction and change log, and a content overview. Collection, media download, and local speech transcription are orchestrated as stdio MCP tools so Codex, WorkBuddy, and other compatible clients can run creator research and scheduled checks without extra API keys.

Local MCP server for agents to search, structure, and export authorized WhatsApp Web conversations. Uses Playwright for DOM interaction, supports message search, export, media transcription, and controlled message sending.

Enables AI agents to fetch and digest content from 30+ platforms (Twitter, YouTube, Reddit, etc.) via a unified API. Supports multi-format output, transcription, and direct Obsidian sync.

A single STDIO MCP server that provides 48 tools for browser and Chrome DevTools automation, HTTP requests, SQL and MongoDB operations, GitBook, system intelligence, disk management, web search, local transcription, and Ollama delegation, plus runtime connection to external auth-free MCP servers.

Local-first MCP server that gives Claude Code web search, page reading, video transcription, and image analysis — without paid API keys. Runs SearXNG + whisper.cpp natively on Apple Silicon for zero-cost, low-latency research workflows.

Enables end-to-end testing of voice-enabled web applications by simulating a synthetic user with a virtual microphone, virtual speakers, a real browser, and a viewable display, allowing audio injection, speech capture/transcription, and browser automation.

Examines web pages through screen-reader actions and speech, enabling agents to perform accessibility checks and interact with page content.

MCP server for extracting links from text dumps, checking duplicates, extracting YouTube transcripts (with fallbacks including speech-to-text), and preparing workspace-ready page payloads for your note-taking or saving tools. It does not write to your workspace; it hands off prepared content to your existing tools.

Turns a YAML script into a narrated product tour video: a real browser follows the steps with a guided focus, captions and text-to-speech, and renders an MP4. Agents can read the script format, validate a script, dry-run it against the real site and render the video.

Lets an AI agent read public TikTok data with no API keys or developer account: search videos and hashtags, fetch profiles, recent videos and comments, discover creators posting about a topic, and transcribe videos through your own OpenAI-compatible speech-to-text endpoint. All popular MCP clients can launch it via uvx, and every tool returns structured output as a read-only operation.