Frequently Asked Questions

No. All audio capture and speech recognition happen locally on your Mac using an offline speech model. No audio is ever transmitted to or stored on any server.

No. Transcripts exist only in memory while the app is open. Nothing is archived to your hard drive or uploaded anywhere. See our Privacy Policy for full detail.

No. TerpScribe does not create or save an audio recording of any proceeding. It displays a live, on-screen text reference as speech happens, and that reference disappears when you close the app.

No. TerpScribe is a personal reference tool for the interpreter's own use during consecutive interpretation. It is not a certified or official record of any proceeding. See our Terms of Use for full detail.

TerpScribe uses an offline automatic speech recognition model and, like any speech-to-text system, its output can contain errors — particularly with uncommon names, heavy accents, or overlapping speech. It's designed as a quick visual aid, not a verified record, and you should always rely on your own judgment over what's on screen.

The app highlights names, addresses, dates, and emails as it detects them, but detection is heuristic and not guaranteed to be correct. Always verify anything you rely on independently. However, if there are specific key names, figures, or terminology you know will come up, you can manually add and edit your own quick reference terms directly inside the sidebar for instant visual access.

No, TerpScribe is currently macOS-only, and specifically requires an Apple Silicon Mac (M1/M2/M3/M4/M5 or newer) — the offline speech model relies on Apple's on-device machine learning framework, which isn't available on Intel Macs or other platforms.

TerpScribe runs 100% locally and requires the following specifications:

• Processor: Apple Silicon M-series chip (M1, M2, M3, M4, M5, or newer). Intel-based Macs are not supported.
• Operating System: macOS 13 (Ventura), macOS 14 (Sonoma), macOS 15 (Sequoia), or newer.
• Memory (RAM): 8GB of Unified Memory minimum (16GB or more recommended for optimal extraction speed).
• Disk Space: 1.5 GB of free space to download and house the local machine learning models.

No. Transcription itself works fully offline. An internet connection is only needed for signing in and verifying your subscription.

TerpScribe features a built-in setup wizard that automates most of the configuration for you. However, you will still need a free virtual audio driver called BlackHole to route the call's audio into the app. The setup wizard inside the app will guide you through installing and configuring it step-by-step, and we also have a detailed setup guide if you prefer manual configuration.

TerpScribe supports 21 languages with automatic language detection: English, Spanish, French, German, Italian, Portuguese, Russian, Dutch, Polish, Czech, Swedish, Danish, Finnish, Greek, Turkish, Arabic, Hindi, Japanese, Korean, Thai, and Romanian. The app can automatically follow a conversation that switches between two configured languages without manual input.

Yes, new accounts include a 14-day free trial period before a subscription is required.

Yes. Subscriptions can be cancelled at any time; billing is handled by Paddle, our merchant of record.

No. Paddle processes and stores all payment details directly. We never receive or store your card number.