All recognition runs on your computer. The audio never leaves it, there is no account and no subscription. Buy a licence for a year or forever.
Apple Silicon (M1 or newer), macOS 13+; system audio needs 14.4. The 700 MB download includes the model. Notarized by Apple, with a free 7-day trial.
The text appears immediately in the active field of any app and is saved to the clipboard.
Double-tap the right Command key during a call in Zoom, Meet, Teams, WhatsApp, Telegram or FaceTime. ShepTone records your voice and the other person’s voice at the same time. It also offers to start recording as soon as it notices a call beginning.
Recognition runs during the conversation, not after it. The transcript has timestamps and speaker separation. The finished text is saved in your Documents folder.
ShepTone copies the transcript together with a ready-made prompt: one for a short summary, one to clean up the text. Paste it into ChatGPT, Claude or DeepSeek.
You decide whether to send the text to a cloud service. Nothing is transmitted without your action.
Otter, Granola, Fireflies and the cloud Whisper API send your conversations to their servers and charge a subscription for it. ShepTone sends nothing: no audio, no text, no telemetry. It does not ask for an account.
Audio, text and telemetry are not sent over the network.
ShepTone is suitable for working with confidential conversations, including in law, medicine, finance and HR.
Russian, English, Spanish, French, Italian, German, Portuguese, Greek — recognition and interface alike.
Choose your main language. ShepTone detects the rest automatically.
Long calls are transcribed without losing the tail of the conversation.
If something goes wrong mid-call, what was already recognized is kept.
When a call starts, ShepTone reminds you to record it.
Violet, Forest and Olive.
Dictation goes into the text field you are typing in.
Summary or clean-up, copied together with the transcript.
Elsewhere those are two separate apps and two separate purchases. Here one licence covers both.
The recognition model ships inside the app, so there is nothing to connect to and nothing to upload.
No subscription, no account, no card on file. The key is verified on your device. There are no automatic charges.
There is no account and no server logs — nothing that could be requested, subpoenaed or leaked later.
| ShepTone | Cloud transcribersOtter, Fireflies, Granola | Other offline appsMacWhisper, VoiceInk | Audio recordersPiezo, Audio Hijack | |
|---|---|---|---|---|
| Works fully offline | ✓ | — | ✓ | ✓ |
| Dictation into any app | ✓ | — | ✓ | — |
| Records both sides of a call | ✓ | ✓ | varies | ✓ |
| Transcript with speakers and timestamps | ✓ | ✓ | varies | — |
| Works without an account | ✓ | — | ✓ | ✓ |
| One-time purchase | ✓ | subscription | varies | ✓ |
See the detailed comparison of ShepTone with MacWhisper, superwhisper, Wispr Flow, VoiceInk and Otter on a separate page.
Both options are paid once. There is no automatic renewal and no card details are stored. The licence is verified on your device, so no internet connection is required.
Try everything for seven days first — both dictation and call recording, all eight languages, no feature held back.
One payment. Nothing auto-renews.
Your email is used for the receipt and the key. Nothing else.
No. Recognition runs on your computer, the model is inside the app, and the licence is checked offline.
Otter and Wispr Flow send audio to their servers and charge a subscription. MacWhisper and VoiceInk work offline too. The difference is that ShepTone combines dictation and recording of both sides of a call, and you pay once. The details are in the full comparison.
It arrives by email within a minute of purchase. Lost the email? Get it re-sent at sheptone.app/key.
The app asks for a new key. Nothing is charged automatically — you decide whether to buy again. Your transcripts are plain files and stay yours either way.
In ~/Documents/ShepTone Transcripts/. They are ordinary files you can move, delete or back up yourself.
Microphone, to hear you. Accessibility, to insert text into the active field and to recognise the key combinations. Screen & System Audio Recording, to capture the other side of a call. Documents folder, to save transcripts. Nothing is requested that is not used for these four things.
No. ShepTone needs Apple Silicon (M1 or newer) and macOS 13 or later; system-audio recording needs macOS 14.4.
About 700 MB, most of it the 574 MB recognition model shipped inside the app.
Yes. Enter the same key on the new Mac — you do not need to unlink the licence from the old device.
The seven-day trial is there so you can decide before paying. If something still does not work for you, write to support@sheptone.app and we’ll refund you.