Logo
Log In
Logo

The Voice AI Platform Most Focused on Developers

ISO 27001
ISO 27001
SOC 2
SOC 2
SSL/TLS
SSL/TLS
APPI
APPI
Products
  • Streaming Speech-to-Text
  • Pre-recorded Speech-to-Text
  • Text-to-Speech
  • Pronunciation Assessment
  • DolphinTeams Dual-Screen Terminal
  • Tralingo AI Translator
  • NihongoScore
Resources
  • Docs
  • Blog
  • AI Apps
  • API Playground
Company
  • About
  • Contact
  • Customers
Legal
  • Privacy Policy
  • Terms of Service
  • Service Level Agreement (SLA)
  • Notations based on SCTA
  • DolphinTeams User Manual
© 2026 DolphinVoice All Rights Reserved.
Productstrnscrb
trnscrb

trnscrb

On-device transcription for macOS with MCP

63 VotesVisit Website
trnscrb Screenshot
trnscrb Preview
Visit Website

Introduction to trnscrb

trnscrb is a macOS application designed for local meeting transcription, offering a private and secure way to capture audio from meetings without relying on cloud services. It runs in the menu bar and automatically detects popular meeting platforms such as Zoom, Google Meet, Microsoft Teams, Slack, and FaceTime. Using the Whisper model on Apple Silicon, it performs on-device transcription, ensuring that no data is sent to external servers. Transcripts are saved as plain text files with speaker diarization, making them easy to review and manage.

The integration with Claude Desktop enhances the user experience by enabling search, summarization, and extraction of action items from past meetings. This makes trnscrb an ideal tool for professionals who value privacy, control over their data, and seamless workflow integration. With no need for an account or subscription, trnscrb provides a straightforward solution for meeting transcription on macOS.

Takeaways

  • Local, on-device transcription using Whisper on Apple Silicon
  • No cloud, no account, no subscription required
  • Supports Zoom, Google Meet, Microsoft Teams, Slack, and FaceTime
  • Transcripts saved as plain .txt files with speaker labels
  • Integration with Claude Desktop for advanced analysis
  • Works offline except for enrich command
  • Automatically detects meeting apps via mic activity and process presence

How trnscrb Works

  1. Auto-detects your meeting: Detects when a meeting starts based on mic activity and running meeting apps.
  2. Records locally: Captures either just your microphone or both mic and system audio via BlackHole virtual driver.
  3. Transcribes on-device: Uses the Whisper small model on Apple Silicon Metal for fast, offline transcription.
  4. Saved to Claude Desktop: Transcripts are stored in ~/meeting-notes/ and can be searched and analyzed through MCP tools.

Core Benefits and Applications

  • Ideal for users who prioritize data privacy and security
  • Useful for teams that want to quickly extract key insights from meetings
  • Enhances productivity by allowing search and summary of past meetings
  • Suitable for remote work environments where internet access may be limited
  • Offers flexibility with audio recording options (mic only or system audio)

MCP Tools

ToolDescription
start_recordingStart capturing audio
stop_recordingStop and transcribe in background
recording_statusCheck if recording or transcribing
get_last_transcriptFetch the most recent transcript
list_transcriptsList all saved meetings
get_transcriptRead a specific transcript
get_calendar_contextCurrent or upcoming calendar event
enrich_transcriptAdd summary + action items via Claude API

Tags

#transcription#macOS#local#privacy#Whisper#Claude#Zoom#Meet#Teams#Slack

Featured

Guideflow

Guideflow

The AI demo automation platform for SaaS

1259
CyberCut AI

CyberCut AI

AI video studio for viral social clips

706
Incredible

Incredible

Deep Work AI Agents - powered by Agent MAX

653
Typeless

Typeless

AI voice dictation that's actually intelligent

625

Showcase your app on AI Apps for free

Join our community of innovators and get your AI tool in front of thousands of daily users.

Get Featured
DolphinVoice Console

BlogPage.PromoContent.title

BlogPage.PromoContent.description

BlogPage.PromoContent.cta