EchoPad iconFree · Open source · Runs on your Mac

Every call, on paper.

EchoPad records your calls and meetings, tells the speakers apart and saves the transcript where you keep your notes. No bot in the call, no cloud, no subscription.

Install EchoPad View on GitHub
EchoPad main window with a transcript split by speaker

How it works

From call to notes without typing

Step 1

Join a call

EchoPad notices Zoom, Meet, Teams and friends using the microphone and offers to record.

Step 2

Talk as usual

Your microphone and the other side are recorded as two tracks. A glass pill shows the timer.

Step 3

Get a transcript

Stop, and a transcript with speakers lands in your folder, in the formats you chose.

Features

A meeting recorder that stays out of the way

Parakeet v3, on-device

NVIDIA’s TDT model runs on Apple’s Neural Engine through FluidAudio. 25 European languages, fully offline after one download.

Knows who said what

Your side comes from the microphone, so it is always you. The other side is split into speakers you can rename.

Saves where you work

An Obsidian vault, a client folder, iCloud Drive. Each save location has its own folder, file name and formats.

Markdown, text, subtitles

Markdown with front matter, plain text, SRT, WebVTT and JSON. Audio as AAC or WAV, next to the note or on its own.

Never miss a meeting

A notification with a Record button appears when a call starts. Or press ⇧⌘E, or click the menu bar icon.

Free, for real

No account, no subscription, no minute limit. MIT licensed, like the two Swift packages it is built on.

A look inside

Conversations, save locations and settings in one window

Save location editor with folder, file name, formats and audio options
Save locations with a live preview of the path
Recording settings: microphone, system audio, meeting detection
What to record, and when to ask

Privacy

What leaves your Mac? Nothing.

  • Audio is recorded only while you record, into a folder on your Mac.
  • Transcription and speaker detection run locally on the Neural Engine.
  • The only network requests download the models from Hugging Face, once.
  • No analytics, no crash reporting, no account. Check the source to be sure.
EchoPadTypical meeting note-taker
PriceFree foreverMonthly subscription
Where audio is processedOn your MacRemote servers
Bot joins the callNoOften
Where transcripts goAny folder you pickTheir web app
Source codeOpen, MITClosed

Open source

Two Swift packages you can use too

ScribeKit

Parakeet v3 transcription with speakers in a few lines of Swift, plus Markdown, SRT, WebVTT and JSON output.

SystemAudioKit

Record the microphone and what the Mac plays, as two aligned tracks, through Core Audio process taps.

Keep every conversation

Build it from source in one command. Setup walks you through permissions, the model download and where to save.

Install EchoPad View on GitHub