trnscrb: Free, Open-Source Local macOS Meeting Transcription with Direct Claude Chat

2026-02-21 | ProductHunt | Official Site | GitHub

Transcription controls are right in your menu bar: Start/Stop, Auto-detect toggle, and open notes folder. It's so simple there's almost no learning curve.

30-Second Quick Take

What is it?: A meeting transcription tool that lives in your Mac's menu bar. It auto-detects Zoom/Meet/Teams meetings, uses Whisper to transcribe locally, and saves .txt files on your computer. The kicker? It has a built-in MCP server, allowing Claude Desktop to search, summarize, and extract action items directly from your transcripts.

Is it worth it?: If you're an Apple Silicon Mac user who attends frequent meetings and hates uploading audio to the cloud, it's worth a shot. It's free, open-source, and installs with three commands. Note: It's a "Day 1" product (launched less than 24 hours ago), so stability is still unproven.

Three Key Questions

Is it for me?

• Target Audience: Knowledge workers on Mac (M1+) who attend remote meetings and prioritize privacy.
• Are you the one?: If you have 3+ Zoom/Meet/Teams meetings a week, often struggle to remember "who said what," and don't want to pay Otter.ai $17/month—you are the target.
• Use Cases:
  • Remote team syncs → Auto-transcribe and let Claude extract action items.
  • Client interviews/User research → Local transcription ensures sensitive data never leaves your machine.
  • 1:1 meetings → Quickly review discussions without taking manual notes.
  • Note: If you only use mobile or Windows for meetings, this isn't for you.

Is it useful?

ROI Judgment: If you already have an Apple Silicon Mac and Claude Desktop, the cost of trying it is nearly zero. However, if you need speaker diarization, multi-language support, or team collaboration features, this tool is currently too early-stage.

What's to love?

The "Wow" Factor:

• Zero-config Auto-detection: Just open Zoom and it starts recording. No more forgetting to hit "Start."
• MCP Integration: Likely the first tool to bridge local transcription with Claude Desktop. Ask Claude "What were today's action items?" without any copy-pasting.

Real User Feedback:

"trnscrb is now live on Product Hunt! A lightweight macOS app for local meeting transcription. Saves clean .txt files on your machine and exposes them via a local MCP server." — @ajayrmk (Developer)

"is the one a breakthrough or a bust?!" — @Looking4Oranges

To be honest, with less than a day since launch, there are very few real-world reviews yet. This is a classic "Day 1 product"—great concept, but needs time to prove itself.

For Independent Developers

Tech Stack

• Desktop: macOS native menu bar app (likely Swift, distributed via Homebrew)
• Speech Recognition: OpenAI Whisper (whisper.cpp) with Apple Silicon Metal GPU acceleration
• MCP Server: Local MCP service exposing .txt transcripts to Claude Desktop
• Distribution: Homebrew tap (brew tap ajayrmk/tap), bypassing the App Store
• Output: Plain .txt files stored locally

Core Implementation

The logic is straightforward: Listen to system audio → Detect meeting apps (Zoom/Meet/Teams/Slack/FaceTime) → Use whisper.cpp for local STT → Save as .txt → Use MCP server to let Claude read the files.

Using whisper.cpp with Metal acceleration on Apple Silicon allows transcription speeds 3-4x faster than real-time. A one-hour meeting can be processed in just a few minutes.

Open Source Status

• Open Source: Yes, GitHub ajayrmk/trnscrb
• Similar Projects:
  • mlx-whisper-mcp — Pure MCP whisper server, no auto-detection.
  • Meetily — Rust-based transcription + Ollama summarization.
  • Notes4Me — BlackHole + Whisper + Ollama local solution.
• Build Difficulty: Low-Medium. The core components (whisper.cpp + MCP SDK) are ready-made. The challenge lies in reliable system audio capture and memory management during long recordings.

Business Model

• Monetization: None; purely free and open-source.
• Pricing: $0
• User Base: Just launched; 82 votes on PH.

Giant Risk

Big players are already here: Zoom AI Companion, Teams Copilot, and Google Meet's built-in transcription. However, these are all cloud-based. trnscrb's edge is 100% local + MCP integration. The real threat is Apple—if macOS adds native local meeting transcription (similar to iOS call recording), these third-party tools may become obsolete.

For Product Managers

Pain Point Analysis

• Problem: Retaining remote meeting content. People forget discussions, and manual note-taking is distracting.
• Frequency/Severity: Medium-high frequency. Managers and heavy meeting-goers need this, though many already use Otter/tl;dv.
• Unique Angle: Privacy + MCP. It's not just a better transcription tool; it's a solution that requires zero trust in third parties.

User Persona

• Primary: Privacy-conscious techies (developers, security researchers, crypto professionals).
• Secondary: Medical/Legal professionals (high compliance needs), freelancers (avoiding subscriptions).
• Scenario: Remote interviews, client calls, or internal strategy sessions where cloud uploads are prohibited.

Feature Breakdown

Competitive Landscape

For Tech Bloggers

Founder Story

• Founder: Ajay Ram K (@ajayrmk), based in Kerala/Bangalore, India.
• Background: Full-stack developer at Sentient Pi with a diverse GitHub portfolio (MERN, Python, Flutter).
• Motivation: A classic "scratch your own itch" project—he wanted local transcription without cloud uploads, integrated with Claude for analysis.

Discussion Angles

• "Is MCP the next killer app?": trnscrb is one of the first to merge MCP with local transcription. If the MCP ecosystem takes off, this is a pioneer.
• "Local vs. Cloud": The eternal trade-off between privacy and convenience. trnscrb goes full local—but that means no speaker ID or team sync.
• "Homebrew vs. App Store": Why skip the App Store? The developer chose a higher-barrier but more flexible distribution method.

Content Suggestions

• Angle: "The MCP ecosystem is birthing a new breed of local tools"—use trnscrb as a case study for the local AI + MCP trend.
• Note: Since the product is very new, it's best featured in a roundup of "Privacy-First AI Tools" rather than a standalone deep dive.

For Early Adopters

Pricing Analysis

Getting Started

• Setup Time: 5 minutes
• Learning Curve: Low (if you know Homebrew)
• Steps:
  1. Open Terminal: brew tap ajayrmk/tap
  2. Run: brew install trnscrb
  3. Run: trnscrb install
  4. Click the menu bar icon to start.
  5. (Optional) Configure the MCP server in Claude Desktop.

Pitfalls & Notes

1. Apple Silicon only: Intel Macs are not supported due to Metal GPU requirements.
2. No App Store: Homebrew installation might be a hurdle for non-technical users.
3. Day 1 Product: No user reports yet. Stability during 2-hour+ meetings and memory usage are unknown.
4. No Speaker ID: All dialogue is lumped together; it doesn't distinguish who is speaking.
5. Hard to Search: Searching "trnscrb" on Google often defaults to other products.

Security & Privacy

• Data Storage: 100% local .txt files.
• Network: Transcription is fully offline.
• Risk: The local MCP server theoretically only allows Claude Desktop to access your files.

For Investors

Market Analysis

• Market Size: AI transcription market was $4.5B in 2024, expected to hit $19.2B by 2034.
• Drivers: Remote work normalization, stricter privacy laws (GDPR), and Apple Silicon enabling local inference.

Timing

• Why now?: Anthropic released the MCP protocol in late 2025; the ecosystem is expanding fast in early 2026. trnscrb occupies the "Local + MCP" niche.
• Market Readiness: Privacy concerns are at an all-time high, though "local transcription" remains a niche compared to the convenience of cloud tools.

Conclusion

trnscrb is a fascinating "Proof of Concept"—it proves the Local + MCP + Claude workflow is viable, but it's far from a mature product.

Resource Links

2026-02-21 | Trend-Tracker v7.3