trnscrb is a macOS menu bar app that automatically detects and transcribes meetings locally using Whisper on Apple Silicon. It saves transcripts as plain text files and integrates with Claude Desktop for searchable meeting notes.

trnscrb is a dedicated macOS application designed to streamline the process of capturing and transcribing meetings entirely on your local machine, offering a private and efficient solution for professionals who frequently participate in virtual discussions. It operates discreetly from the menu bar, automatically identifying when you are in a meeting by monitoring microphone activity and the presence of specific communication applications like Google Meet, Zoom, Slack, Microsoft Teams, or FaceTime. The primary purpose of trnscrb is to provide users with accurate, searchable transcripts of their conversations without ever needing to send sensitive audio data to the cloud, thereby ensuring complete privacy and data ownership while integrating seamlessly with tools like Claude Desktop for enhanced note-taking and retrieval.

In today's remote and hybrid work environments, professionals are inundated with back-to-back virtual meetings, leading to information overload and the constant struggle to recall specific details, decisions, or action items discussed. Manually taking notes during these sessions is distracting and often incomplete, while relying on memory alone is unreliable and inefficient. The common pain point is the loss of critical information buried in hours of meeting recordings or the lack of a quick, private way to review what was said, who said it, and what commitments were made, creating bottlenecks in project follow-through and collaborative accountability.

The first major feature group is its fully automatic meeting detection and local recording capability. trnscrb intelligently detects when a meeting begins by checking if the microphone has been active for five seconds and if a supported meeting application such as Zoom, Google Meet, Slack Huddle, Microsoft Teams, or FaceTime is running on the system. Once detected, it immediately starts recording audio locally, capturing either just the user's microphone input or, when configured with the BlackHole virtual audio driver, both the microphone and the system audio to include other participants' voices. This local-first approach means no audio data leaves the computer during capture, addressing privacy concerns from the outset and ensuring recordings are always available even without an internet connection.

The second major feature group is its on-device transcription powered by OpenAI's Whisper model, specifically optimized for Apple Silicon Macs. After a recording session ends, trnscrb processes the audio file locally using the Whisper `small` model, which runs efficiently on Apple's Metal framework for hardware acceleration, delivering fast and accurate transcriptions without requiring any internet connectivity. This offline transcription capability is a cornerstone of the app's privacy promise, as it guarantees that sensitive conversation content is never uploaded to external servers, with the only exception being an optional enrichment step that users must explicitly initiate. The transcripts are saved as plain text files in a designated folder, making them simple to access and process.

trnscrb

Key Features

Use Cases

Who is this for?

Comments