Trace - Offline Mac Meeting Transcripts

TL;DR

TL;DR: Trace is a $9.99 macOS menu-bar app that transcribes meetings entirely on-device using Apple silicon, outputs clean markdown with inline flagged moments, and never sends audio to the cloud.

Source and Accuracy Notes

• Project page: traceapp.info
• Mac App Store: Trace - On-Device Transcripts
• Price: $9.99 (verified from App Store metadata)
• HN launch thread: news.ycombinator.com/item?id=48521236 (201 points, 81 comments)
• Source last checked: 2026-06-16

What Is Trace?

Trace is a macOS menu-bar app that captures your microphone and system audio, transcribes conversations locally on your Mac, and produces clean markdown transcripts with flagged moments inline. It runs a local speech model entirely on-device - no cloud services, no accounts, no meeting bots joining your calls.

The app works over any conferencing tool: Zoom, Microsoft Teams, Google Meet, Slack, or anything else that produces audio on your Mac.

Key Features

On-Device Processing

Trace runs a local speech model on Apple silicon. According to the developer, a typical meeting transcribes in seconds rather than minutes. No upload queue, no cloud spinner.

“A local speech model, running on-device. No servers, no accounts, and nothing you record ever leaves your Mac.”

Key Moment Flagging

During a call, press a keyboard shortcut to flag important moments with a short note. The flag lands in your transcript at the exact timestamp:

• Record: Cmd+R
• Flag key moment: Cmd+K (type a note, hit Enter)
• Recap: Cmd+?
• Pause: Cmd+P
• Hide pill: Cmd+H

All shortcuts are remappable.

Clean Markdown Output

Each session produces files stored in ~/Application Support/Trace/:

• transcript.md - the main markdown transcript
• transcript.json - structured JSON version
• meta.json - session metadata
• mic.wav - your microphone audio
• system.wav - system/call audio

The markdown is designed to paste directly into Notion, Obsidian, ChatGPT, Claude, or any tool that accepts markdown.

Menu Bar Interface

Trace lives in your menu bar and stays out of the way until you need it. One keystroke brings it up over any application. The menu bar shows a list of recent sessions with copy, reveal in Finder, or re-open options.

Setup and Requirements

System Requirements

• macOS: 14.4 or later
• Hardware: Apple silicon (M1, M2, M3, M4)
• Price: $9.99 (one-time purchase on Mac App Store)

Installation

1. Download from the Mac App Store
2. Launch Trace - it appears in your menu bar
3. Grant microphone access when prompted
4. Press Cmd+R to start recording during any call

First Recording

When you start a recording, Trace captures both your microphone input and system audio (the other participants). When you stop, it transcribes locally and saves the session files.

Privacy Model

Trace’s privacy proposition is straightforward: audio never leaves your Mac. There is no server component, no account system, and no telemetry mentioned on the website. The speech model runs locally on Apple silicon’s Neural Engine.

This contrasts with cloud-based transcription services (Otter.ai, Fireflies, Grain) that upload audio to remote servers and require meeting bots to join your calls.

Deeper Analysis

Who Is This For?

Trace targets developers, founders, and knowledge workers who:

• Attend many video calls and need transcripts for reference
• Want transcripts without inviting a third-party bot to the meeting
• Use Apple silicon Macs and value on-device processing
• Work with markdown-based workflows (Obsidian, Notion, plain text)
• Need to flag specific moments during long meetings for later review

Comparison with Alternatives

Cloud transcription services (Otter, Fireflies, Grain):

• Require meeting bots or audio upload
• Monthly subscriptions ($10-25/month)
• Speaker diarization and AI summaries included
• Audio stored on third-party servers

Trace:

• One-time $9.99 purchase
• No cloud, no accounts
• Markdown output you own completely
• No AI summaries (you paste into ChatGPT/Claude yourself)
• Requires Apple silicon

Limitations

• Apple silicon only (no Intel Mac support)
• macOS 14.4+ required
• No built-in speaker diarization mentioned (unclear if it distinguishes speakers)
• No AI summarization built in (manual paste to LLM)
• English-only transcription (not verified, likely limitation of local model)

Practical Evaluation Checklist

Before buying Trace, verify:

• [ ] You have an Apple silicon Mac (M1/M2/M3/M4)
• [ ] You run macOS 14.4 or later
• [ ] You attend enough video calls to justify $9.99
• [ ] You are comfortable pasting transcripts into AI tools for summaries
• [ ] You do not need automatic speaker diarization
• [ ] You want markdown output, not a proprietary format
• [ ] Privacy (no cloud upload) is a priority for your use case

Security Notes

• Audio processing happens entirely on-device
• No account creation required
• No network calls mentioned in the app description
• Session files stored locally in ~/Application Support/Trace/
• You control deletion via Finder (plain files, not a database)

If your organization has strict data handling requirements, Trace’s on-device model avoids the compliance review that cloud transcription services typically require.

FAQ

Q: Does Trace work with Intel Macs? A: No. Trace requires Apple silicon (M1, M2, M3, M4) and macOS 14.4 or later.

Q: Can Trace distinguish between different speakers? A: The website does not mention speaker diarization. You may need to manually label speakers in the transcript or use an AI tool to add speaker labels after transcription.

Q: Does Trace transcribe in languages other than English? A: The website does not specify language support. The local speech model likely focuses on English, but this is not confirmed.

Q: Can I use Trace for in-person meetings, not just video calls? A: Yes. Trace captures microphone input, so it works for any audio source your Mac can hear, including in-person meetings if you place your Mac near the conversation.

Q: Is there a free trial? A: The App Store listing does not mention a free trial. The $9.99 price is a one-time purchase.

Q: Does Trace integrate with Calendly, Zoom, or other scheduling tools? A: Trace does not integrate with scheduling or conferencing tools directly. You manually start and stop recordings during calls.

Conclusion

Trace fills a specific niche: privacy-conscious meeting transcription for Apple silicon Mac users who want markdown output without cloud dependencies. At $9.99 one-time, it is significantly cheaper than monthly transcription subscriptions, but it lacks features like automatic speaker diarization and AI summaries that cloud services provide.

If you attend many video calls, use markdown-based note-taking tools, and prioritize on-device processing, Trace is worth evaluating. If you need speaker identification, automatic summaries, or cross-platform support, cloud-based alternatives remain the better choice.

The app’s HN launch (201 points, 81 comments) suggests strong interest in privacy-first transcription tools, particularly among developers and founders who want control over their meeting data.