Docs on noScribe

Download and Installation

Mon, 01 Jan 0001 00:00:00 +0000

Download and Installation#

Current version: 0.7

All installation files are hosted on SWITCHdrive.ch, a secure data-sharing platform for Swiss universities. The source code is available on GitHub, also for the noScribe Editor.

Windows Installation macOS Installation Linux Installation

Windows#

Download#

Two versions are available. Most users should choose the standard version. If you have an NVIDIA graphics card (RTX 20XX or newer) with at least 6 GB VRAM, you can benefit from CUDA acceleration, which can drastically reduce transcription time.

Usage

Mon, 01 Jan 0001 00:00:00 +0000

Usage#

Settings#

Choose audio file: Almost all audio and video formats are supported. You can also select multiple files in the dialog; they will then be processed one after another.
Save transcript as: Default is .html (compatible with the editor). Optionally .txt for plain text or .vtt for subtitles and further work in EXMARaLDA. If multiple input files are processed at once, only the output folder can be selected. Output file names are then generated automatically.
Start/Stop: Time range in the format hh:mm:ss, ideal for testing shorter excerpts.
Language: Select a language or “Auto” for detection, or “Multilingual” for multiple languages in the same recording (experimental).
Quality: “Precise” delivers the best results and is the recommended choice for almost all use cases. “Fast” is a bit quicker and mainly intended for older computers. It requires more post-correction.
Mark pause: If selected, pauses are marked as dots in brackets, one dot per second of pause. Optional from 1/2/3 seconds.
Speaker detection: If known, select the number of speakers to make detection more robust. Otherwise choose “Auto” for automatic detection or “None” to skip this step entirely.
Overlapping speech: Simultaneous speech is marked with // Speaker: text of the interjection // (experimental).
Disfluencies: If selected, filler words and incomplete words/sentences are transcribed when possible; otherwise they are not. Note: This is more a “recommendation” to the AI model, not a hard on/off switch.
Timestamps: Inserts a timestamp in the format [hh:mm:ss] every 60 seconds or on speaker changes (e.g., useful for MAXQDA).
Start begins the transcription process. If a process is already running, you can still submit new ones. These are then queued in the queue and processed when it is their turn.

Queue#

The “Queue” tab shows a list of all jobs in the current session, their current status, and progress. Jobs are processed one after another.

Advanced Options

Mon, 01 Jan 0001 00:00:00 +0000

Advanced Options#

Scripting#

Since version 0.7, noScribe has a command-line interface and can be integrated into your own scripts. noScribe.exe --help shows the available options. With --no-gui, the main window is suppressed and noScribe runs fully in command-line mode.

Configuration#

After the first start, the file config.yml is located in the user profile:

Windows: C:\Users\<username>\AppData\Local\noScribe\noScribe\config.yml
macOS: ~/Library/Application Support/noscribe/config.yml
Linux: ~/.config/noScribe/config.yml

There you can adjust the UI language and other options. Note: Only edit this file when noScribe is not running, otherwise your changes will be overwritten.

FAQ

Mon, 01 Jan 0001 00:00:00 +0000

FAQ#

Can I trust that noScribe does not send audio data or transcripts to the outside?#

Yes. noScribe was designed from the start as a purely local desktop app. That was the central motivation for building it at all. If you want to test this, temporarily disable your network connection (turn off Wi-Fi, unplug the LAN cable). noScribe will work exactly as before.

Yes, as far as I can tell. Note, however, that I am not a lawyer. But purely local processing of sensitive data is considered as gold standard for privacy - at least if you protect your own computer well against unauthorized access.

Legal Notice & Privacy

Mon, 01 Jan 0001 00:00:00 +0000

Legal Notice#

Information in accordance with § 5 DDG (Germany)

Kai Dröge
Emil-Claar-Straße 8
60322 Frankfurt am Main
Germany
E-Mail: kai.droege@hslu.ch

Privacy Policy for the Website#

This website uses no cookies and no tracking/analytics.

Controller#

Kai Dröge
Emil-Claar-Straße 8
60322 Frankfurt am Main
Germany
E-Mail: kai.droege@hslu.ch

Hosting (GitHub Pages)#

This website is provided as a static site via GitHub Pages. When you access it, visitor IP addresses are logged and stored by GitHub for security reasons. The recipient of the data is the hosting provider GitHub (GitHub, Inc.). The legal basis is Art. 6(1)(f) GDPR (legitimate interest in a secure and reliable operation of the website).

Docs on noScribe

Download and Installation

Download and Installation#

Windows#

Download#

Usage

Usage#

Settings#

Queue#

Advanced Options

Advanced Options#

Scripting#

Configuration#

FAQ

FAQ#

Can I trust that noScribe does not send audio data or transcripts to the outside?#

Is noScribe a GDPR-compliant transcription solution?#

Legal Notice & Privacy

Legal Notice#

Privacy Policy for the Website#

Controller#

Hosting (GitHub Pages)#