BMLV-RadioDocAI

Speech-to-text tools for radio communication in the Austrian armed forces.

Background

Two-way radios are irreplaceable in military service because conventional mobile telephone networks cannot provide the necessary encryption technologies for safe and secure communication. Furthermore, radios can operate reliably in remote and inaccessible locations. In some missions, radio communication is recorded to document operational events. The current practice of making written records retroactively is both time-consuming and prone to errors. This project investigates if modern speech-to-text systems can automate the documentation process.

Contents and Goals

The main objective of the project is to determine whether open-source speech-to-text systems are suitable for translating radio communication of the Austrian Bundesheer into text. The technical part is covered by the University of Applied Sciences St. Pölten, while AI-Trust ZT GmbH is in charge of evaluating and validating the developed solution. Below are the most significant subgoals and important steps of the project:

  • Speech-to-text demonstrator: Development of an automatic transcription prototype which can cope with Austrian dialects and military jargon.
  • Benchmarking and evaluation: Testing the practicability of various solution approaches.
  • Technical integration: Implementation of a speech-to-text transcription demonstrator.
  • Trial Implementation in Training and Exercise Operations: The prototype is tested under realistic conditions.
  • Further development of AI-models: Training, adjusting and evaluating of the speech-to-text models. The models will be refined incrementally based on the insights gained during the trials.
  • Recommended actions and roadmap for further lines of action: Final evaluation, recommendations and the development of technical guidelines.

Results

We examine which AI models are most suitable for the use of automatically translating military radio communication. We develop a demonstrator and test it under real-life conditions. The insights we gain will shed light on the key challenges speech-to-text models must be able to overcome. Additionally, our project lays the foundation to turn our prototype into a usable and operational tool.

Partner

Funding

You want to know more. Feel free to ask.

Lecturer Industry 4.0
Education Coordinator
Media and Digital Technologies
Course Leader Agricultural Technology (acad.)
Course Leader Certified Professional for UX-Development (certif.)
Course Leader Digital Photography and New Visual media (MA)
Course Leader Production Management (acad.)
Head of Josef Ressel Center for Knowledge-Assisted Visual Analytics for Industrial Manufacturing Data
Department of Media and Digital Technologies
Location: A - Campus-Platz 1
M: +43/676 847 228 689
Partners
  • AI-Trust ZT GmbH
Funding
Bundesministerium für Landesverteidigung (Auftrag)
Runtime
11/01/2026 – 08/31/2027
Status
current
Involved Institutes, Groups and Centers
Institute of Creative\Media/Technologies