trnscrb is an open source, privacy focused meeting transcription tool built for macOS. It automatically detects online meetings, records audio locally and uses OpenAI’s Whisper speech recognition model to create transcripts directly on the user’s computer.
Unlike many AI meeting assistants that upload recordings to cloud servers, trnscrb performs its core recording and transcription locally. This means an internet connection is not required for normal transcription, and meeting audio does not need to leave the user’s Mac.
The tool can detect meetings taking place through Google Meet, Zoom, Microsoft Teams, Slack Huddles and FaceTime. It can capture the user’s microphone alone or both microphone and system audio when configured with the BlackHole virtual audio driver.
Transcripts are stored as ordinary TXT files rather than inside a proprietary cloud database. trnscrb also integrates with Claude Desktop through MCP, allowing users to search previous meetings, retrieve transcripts and work with their meeting information using natural language.
It is particularly suitable for Mac users, developers, professionals and privacy conscious teams that want local meeting transcription without maintaining another cloud based meeting notes subscription.
Features
Automatic Meeting Detection
trnscrb can automatically recognize when the user has entered a supported meeting. Detection is based on microphone activity combined with the presence of applications or services such as Google Meet, Zoom, Teams, Slack or FaceTime.
Offline AI Transcription
The platform uses the Whisper small model to convert speech into text directly on the Mac. On Apple Silicon devices, Whisper can take advantage of Metal acceleration for faster local processing.
Local Audio Recording
Meeting recordings are processed locally instead of automatically being uploaded to external cloud infrastructure.
Google Meet Support
trnscrb can detect and record meetings conducted through Google Meet, including meetings joined through a web browser.
Zoom Support
Zoom calls can be automatically detected and converted into searchable transcripts.
Microsoft Teams Support
The tool supports Microsoft Teams meetings, allowing users to maintain local written records of conversations.
Slack Huddle Support
Slack Huddles are included among the meeting environments detected by trnscrb.
FaceTime Support
Mac users can also use trnscrb with FaceTime conversations.
Microphone and System Audio Capture
Users can record only their microphone or capture both microphone and system audio. Capturing system audio requires the free BlackHole virtual audio driver.
Speaker Labels
Speaker separation can identify participants with labels such as SPEAKER_00 and SPEAKER_01.
A free Hugging Face account is required to enable speaker labels. Transcription itself continues to work without a Hugging Face account if speaker identification is not required.
Plain Text Transcripts
Completed meeting transcripts are stored as standard TXT files inside the user’s meeting-notes directory. Users therefore retain direct access to their files without depending on a proprietary storage format.
Claude Desktop Integration
trnscrb integrates with Claude Desktop using MCP. Claude can access supported local meeting tools, allowing users to work with their transcript collection conversationally.
Search Previous Meetings
Through Claude Desktop integration, users can ask questions about information contained across previous meeting transcripts instead of manually opening and reading every file.
Meeting Summaries and Action Items
An optional enrichment command can send transcript text to the Claude API to create summaries and action items.
This feature is different from the standard transcription process because it requires an internet connection and sends transcript text to the external Claude API.
MCP Tools
trnscrb provides MCP functions for starting and stopping recordings, checking recording status, retrieving the latest transcript, listing transcripts, opening a particular transcript and obtaining calendar context.
Automatic Audio Deletion
According to the official website, audio is processed locally and deleted after transcription. The resulting text transcripts remain stored on the user’s computer.
No Proprietary Database
The tool does not require users to maintain their transcripts inside a dedicated cloud database. Files remain accessible as normal text documents.
Open Source
trnscrb is distributed under the MIT License, giving developers greater transparency and flexibility than a closed source transcription service.
How It Works
First, users install trnscrb on their Mac. The recommended installation method is through Homebrew, although Python based installation using pip or uv is also supported.
The installation process sets up required components, including the Whisper model, BlackHole audio configuration, optional Hugging Face requirements, Claude Desktop MCP configuration and launch at login.
Once installed, trnscrb runs from the macOS menu bar.
When microphone activity continues for approximately five seconds while a supported meeting application is active, the tool can detect the meeting.
trnscrb records the audio locally. Users can capture only their own microphone or configure BlackHole to capture system audio so that other meeting participants are included.
After recording ends, the Whisper model processes the audio directly on the computer.
The resulting transcript is saved as a plain TXT file in the local meeting notes folder.
Users can open these files directly or access them through Claude Desktop.
With Claude integration, users can ask questions about previous meetings, retrieve information or search across multiple transcripts.
If users explicitly run the enrichment feature, transcript text can be sent to the Claude API to generate summaries and action items.
Use Cases
Business Professionals: Professionals can maintain searchable records of meetings without manually taking detailed notes throughout every call.
Remote Teams: Teams using Zoom, Meet, Teams or Slack can create transcripts of remote discussions.
Developers: Technical users can benefit from the open source architecture, MCP integration and local storage model.
Privacy Conscious Users: People who do not want meeting recordings automatically uploaded to a transcription company’s servers can process them locally.
Managers: Managers can search previous conversations to locate decisions, commitments and important discussion points.
Project Teams: Project meetings and standups can be transcribed so important decisions remain available for later reference.
Researchers: Interviews and discussions conducted through supported meeting applications can be converted into locally stored text.
Consultants: Client conversations can be transcribed and reviewed after meetings, subject to appropriate recording consent and confidentiality requirements.
Students: Online classes, discussions and study meetings can be converted into text for later review where recording is permitted.
Claude Desktop Users: People already using Claude Desktop can turn their collection of local meeting transcripts into a searchable conversational knowledge source.
Pricing
trnscrb itself is free and open source under the MIT License. The official website does not list a paid subscription for its core local recording and transcription functionality.
Because transcription runs locally with Whisper, users do not need to pay per transcription minute to trnscrb.
Speaker identification requires a Hugging Face account, which the official website describes as free.
The optional trnscrb enrich capability uses the Claude API to generate summaries and action items. Using an external API may involve separate charges from the relevant service provider.
Users therefore should distinguish between the free local transcription capabilities of trnscrb and optional external AI services that may have their own usage costs.
Strengths
Privacy is one of trnscrb’s most important strengths. Core audio recording and transcription take place locally rather than requiring meeting recordings to be sent to a cloud transcription provider.
Another significant advantage is the absence of a recurring subscription for the core tool.
Plain text storage gives users control over their meeting data and avoids locking transcripts inside a proprietary application.
Automatic detection of popular meeting platforms reduces the effort involved in remembering to manually configure a transcription workflow for every call.
Claude Desktop integration is particularly useful because it allows users to search and interact with their locally stored meeting history using natural language.
Open source distribution under the MIT License is another advantage for developers and technically experienced users who value transparency.
Drawbacks
trnscrb is primarily a macOS tool, making it unsuitable for users who mainly work on Windows, Linux, Android or other platforms.
Although it can run on Intel Macs, the official website recommends Apple Silicon because Whisper cannot use Metal acceleration on Intel machines, resulting in slower transcription.
Installation is more technical than many commercial meeting assistants. Users may need to work with Homebrew or Python and understand components such as BlackHole, Hugging Face and MCP.
Capturing both sides of a meeting requires BlackHole system audio configuration.
Speaker identification requires a Hugging Face account.
The core tool focuses more on local transcription than polished cloud collaboration. Teams looking for centralized dashboards, shared workspaces and extensive administrative features may prefer a commercial meeting platform.
Using Claude based enrichment also changes the fully local workflow because transcript text is sent to the Claude API when the user explicitly activates that feature.
Comparison with Other Platforms
Compared with cloud based AI meeting assistants, trnscrb takes a substantially different approach to privacy. Core transcription is processed on the user’s Mac rather than automatically sending meeting audio to cloud servers.
Compared with meeting bots that join calls as visible participants, trnscrb operates locally from the user’s Mac and detects supported meetings based on microphone activity and the applications being used.
Compared with standalone Whisper implementations, trnscrb adds practical meeting functionality such as automatic meeting detection, system audio capture, speaker labels, transcript management and Claude Desktop integration.
Commercial AI meeting assistants may provide more advanced team collaboration, cloud synchronization, shared dashboards and administrative controls. trnscrb instead prioritizes local ownership, simplicity, privacy and open source access.
Its combination of local Whisper transcription and MCP integration with Claude Desktop is particularly useful for technically comfortable Mac users who want AI searchable meeting records without automatically storing their conversations in another cloud platform.
Customer Reviews and Testimonials
Customer reviews and testimonials are not clearly available on the official website.
Public product listings show early interest in the tool, but there is not yet enough established customer review information to provide a meaningful assessment of long term user satisfaction.
Conclusion
trnscrb is a compelling option for Mac users who want private AI meeting transcription without relying on a traditional cloud based meeting assistant.
Its ability to automatically detect meetings, record locally, transcribe with Whisper and store results as ordinary text files creates a simple workflow where users maintain control over their meeting data.
Claude Desktop integration adds another useful layer by allowing users to search and interact with their meeting history through natural language. This can turn a folder of transcripts into a practical source of organizational knowledge.
trnscrb is especially suitable for developers, professionals, researchers and privacy conscious users comfortable with a slightly more technical installation process. For users who value local processing, open source software and ownership of their meeting transcripts more than cloud dashboards and team collaboration features, trnscrb provides an interesting alternative to subscription based AI meeting assistants.
