Skip to content
folo, home

RESOURCES

Local meeting transcription on Windows and Linux

Last updated: November 2026

Most meeting notetakers assume a Mac, a browser or a bot. On Windows you often get a web app or a bot, on Linux usually only the browser. Yet both systems can record a call’s sound themselves, and open speech models run on an ordinary laptop. Here is how local transcription works on Windows and Linux.

What the system can record

A call has two sides: your microphone and the sound the meeting app plays. Both systems let an app record the second one without joining the meeting.

  • Windows has loopback capture in WASAPI: an app records what your speakers play. Newer versions can also capture a single app, say only Teams.
  • Linux with PipeWire or PulseAudio offers a monitor source for every output. Recording it gives you the call’s sound, on X11 and Wayland alike.

Keep the microphone and the call’s sound as two tracks. That makes “who said what” much easier later.

Which models write it down

Two open models cover most needs and run without a graphics card:

  • NVIDIA Parakeet (TDT 0.6B v3): 25 European languages, small and fast.
  • Whisper large-v3-turbo: 99 languages, for everything else.

For summaries and to-dos you need a language model too. That can stay local as well: Ollama or a local server such as LM Studio or llama.cpp.

Cloud notetaker or local app

A cloud notetaker gives you shared notes, web and phone access and nothing to install. In return your audio and transcripts live on its servers, often with a bot in the call, and you pay per person and month. A local app keeps the recording on your disk and works offline. You give up the shared notebook.

If your notes belong to a team or must be on your phone, a cloud tool fits better. If they are yours alone, local is simpler. folo vs Otter.ai compares the two.

folo on Windows and Linux

folo runs on Windows 10 and 11 (64-bit) and on Linux x86_64 with X11 or Wayland, the same as on the Mac. It sits in the tray and asks to record when Zoom, Teams, Webex, Slack or Discord takes the microphone. Meet and Teams in the browser it notices if you switch on Detect video calls in the browser. It records microphone and call sound as two tracks, transcribes with Parakeet or Whisper on your computer, tells speakers apart and writes a summary with to-dos. Voice notes (hold Ctrl+Alt+N) and dictation into any app (hold Ctrl+Alt+X) are included.

On Windows you get an installer, on Linux an AppImage, and both update from inside the app. The Windows installer isn’t code-signed yet, so SmartScreen may warn you: click More info, then Run anyway. Try it free for 14 days.

Questions

Is there an Otter app for Linux?

Otter’s desktop app is for Mac and Windows. On Linux you use Otter in the browser or let its notetaker join. Either way the audio is transcribed on Otter’s servers.

Do I need a graphics card for local transcription?

No. Parakeet and Whisper run on an ordinary laptop. A faster computer just finishes sooner.

Does local recording work on Wayland?

Yes. Audio goes through PipeWire or PulseAudio, the same on X11 and Wayland. Global shortcuts are different: on Wayland your desktop has to allow them, and some sessions don’t.

Read on