Earkeep
Private by design
Earkeep's audio never leaves your computer. It is processed only in memory, transcribed on-device, and never written to disk or sent to any server, API, or account. What it leaves behind is plain files on your own disk that you own outright.
The problem: most "transcription" still means "upload"
- Cloud notetakers join your call as a bot and keep the recording and transcript on their servers, under a privacy policy you didn't write.
- "Local" apps that still call out run on your machine but send the audio to a cloud API (OpenAI's Whisper API, Groq, or similar) for the transcribing. Your recording still leaves the building.
Earkeep does neither. There is no bot, no upload step, and no third-party API in the transcription path.
The second half shows up later: most notetakers keep your history behind their app and their subscription. Stop paying and you typically lose access to it, and export, if it exists, is a partial copy.
How the audio side works, mechanism by mechanism
Audio lives in memory only
Mic and system audio are captured into memory and never written to disk in any form, raw or intermediate: no cached audio file, no temp .wav. Once a chunk is transcribed, the audio behind it is gone.
Transcription runs on-device
Earkeep transcribes with Whisper or Parakeet models running locally, hardware-accelerated on Apple Silicon: the same class of model cloud services run, on your own hardware.
No upload, no API, no server, no account
No upload step, no API call carrying audio to a third party, no transcription server, and no sign-in to use the app. Disconnect from the internet and transcription keeps working.
No bot joins your calls
Earkeep adds no participant to your calls and needs no calendar app authorized to "listen in." It captures system audio like any other audio playing on your computer, with nothing for others to notice on their end.
License checks don't phone home on every launch
License validation uses an offline grace period: the app checks in periodically and keeps working through normal offline stretches in between. It has nothing to do with your meetings.
What gets saved: plain files, on your disk
Earkeep writes one JSONL file per day (one JSON object per line), appended to as it listens. The meetings you keep are marked inside that file where each begins and ends, not copied out, so a correction shows everywhere at once and deleting a meeting leaves every word where it was.
- Any text editor opens them. No proprietary viewer.
- Search them with what you already use, from your editor's find to the command line.
- Any backup tool already handles them, including a private Git repository (Earkeep can make those commits itself).
- Agents and scripts can read them directly. See agents and MCP.
There's nothing to export, because nothing was ever locked in.
What this means if you're thinking about GDPR
Your audio is processed only on your device and never transmitted to us or to any third-party transcription API, so there's no processor relationship for that audio and nothing for us to hold that could be the subject of an access or deletion request. Your transcripts live in files under your control. The data that does involve us, purchase and support email, is in the privacy policy.
On-device processing does not change what you owe the other people in the conversation. In the EU that means a lawful basis for recording (at work, usually legitimate interest) and telling participants beforehand, as Articles 13 and 14 ask. In the United States most states need one party's consent, which can be yours, while a minority require everyone's; published lists of which run from eleven to fourteen, so a call across state lines is safest under the strictest rule on it.
In practice that is a sentence at the top of the meeting; telling people you're recording has the wording. What on-device transcription removes is everything after it: no third-party recipient to name, no processor agreement, no transfer out of the country, no vendor retention policy. Your employer may have its own rules, and this is a description of how Earkeep works, not legal advice.
Related
- Always-on capture: how the transcript gets there, without a record button.
- Your agents, right where your meetings are: agents on transcripts that never left your machine.
Frequently asked questions
Does audio ever leave my computer?
No. Audio is captured into memory for transcription and never written to disk, uploaded, or sent to any server or API.
What format exactly are my transcripts saved in?
One JSONL file per day, one JSON object per line, with the meetings you keep marked inside it. Plain text: no database, nothing that needs Earkeep to open.
Can I export my meetings?
There's nothing to export: your transcripts are plain files from the moment Earkeep writes them.
Can other tools read my transcript files?
Yes. Anything that reads text or JSON: a text editor, grep, jq, a script, a coding agent.
How do I back them up?
With whatever already backs up your files. They're ordinary files, so Earkeep needs no backup mechanism of its own.
What happens to my files if Earkeep disappears?
They stay exactly as readable: plain text doesn't stop working when the app that wrote it does. And per our update policy, the app itself keeps working without future updates.
What data does the app or website send at all?
Nothing from your meetings leaves the device. The app validates its license periodically and sends usage statistics unless you switch them off in Settings → General. Anything else, like connecting a calendar, is your choice.
How exactly does the license check work?
Earkeep validates your license key periodically, not on every launch.
What happens if I use Earkeep entirely offline?
Transcription never needs a network. License validation tolerates normal offline stretches; only an unusually long one would eventually require reconnecting to re-validate.
Your audio never leaves, your files stay yours
14-day full trial, no account required.