A notebook that listens

Say it out loud. Keep the notes, not a cloud copy.

Record an appointment, meeting or passing thought. Turn your words into searchable text and well-organized documents, right on your own computer.

No account, no subscription, no recording uploads with local models

ScribeWatch

Real app screenshot, demo content only — never patient data

01 / Why it matters

Your voice stays yours.

With the complete edition and its locally installed models, your recordings and text are processed on your device. They are not sent to a remote service for transcription or writing.

You need Internet once to download the models. After that, local processing works offline. If you intentionally configure an external provider in advanced settings, that provider’s privacy terms apply.

02 / How it works

From speaking to keeping.

Three simple steps to turn a conversation into something you can use.

  1. 01

    Record

    Press REC, speak, and stop. Or import an audio file you already have.

  2. 02

    Get the words

    ScribeWatch transcribes the audio into text and preserves the original transcript.

  3. 03

    Take away a document

    Pick a note type: consultation, meeting, ideas or plain transcript. Export as PDF, Word or text.

For the things you actually do.

Appointments

Keep the points covered in a conversation and draft a structured report. A professional must review medical reports.

Voice memos

Speak your thoughts. Get editable text you can find and organize later.

Meetings and interviews

Keep a readable record of discussions, decisions and follow-up actions.

04 / Download

Ready to use it?

Pick your computer. Most people should choose the complete edition: it includes everything needed to transcribe and write locally.

Complete edition

Record, transcribe and create documents locally. The required models download on first launch.

v0.6.2
Official downloads from the latest stable release

If the complete build for your system is not yet listed in the latest release, see the advanced editions below.

Already running your own speech engine? Lightweight builds for advanced users

These editions do not install a transcription or writing engine. They require an existing compatible service you have already configured and are not recommended for first-time users.

Browse all GitHub releases

05 / What does your computer need?

What does your computer need?

No subscription or separate server for the complete edition. These are practical guidelines, not guaranteed minimums for every system.

To transcribe

A recent 64-bit computer, microphone for live recording and available disk space for the speech model.

To write structured reports

16 GB RAM is recommended for comfortable use of the local writing model. Lower-spec computers may be slower.

To make it faster

A compatible graphics card can speed things up. Without a dedicated GPU, the app can fall back to the processor, which takes longer.

Keep several gigabytes of storage free for the models and their files. Exact requirements depend on your system and selected models. New complete editions include document exporters without installing Pandoc or LibreOffice separately.

06 / FAQ

A few simple answers.

Does my audio go to the Internet?

Not when you use the local models in the complete edition. Audio and text are processed on your device. The first model download needs Internet; external providers remain an optional advanced choice.

Do I need to understand AI?

No. Install the complete edition, let it set up the local models on first launch, then record or import audio.

What if I do not have a powerful graphics card?

The processor can do the work, only more slowly. For structured reports, 16 GB RAM is recommended.

Can I use it for medical consultations?

Yes. There are dental and medical templates. Every generated report is a draft requiring professional review and correction before clinical use.

Does it work without Wi-Fi?

The first run needs Internet to download the models. Once those are installed, local recordings and reports can be processed offline.

Medical records require professional review. Output quality depends on recording quality, language and hardware.
ScribeWatch