Innovation

Research and development across the company — the matrix every division draws on.

Entel is a Hungarian-owned, deep-tech engineering company founded in 1990. We do not run research and software development as a separate division but as a capability spanning the whole company: the same team and toolset serves telecom consulting, AI systems, acoustic design and audio technology products.

This page presents our development areas. Some of the specific projects, publications and lab results are still published on the division sites, linked at the bottom of the page.

How we work

Matrix organisation

Researchers and developers do not belong to a single division. Each project is staffed with the engineers who know the field best plus AI, signal-processing and software competencies.

From research to product

Every development aims at a working service or product: Soundy, ÁTMR, irat.ai and the Inspired Acoustics products all grew out of internal R&D.

Our own tools

What we build for ourselves we also offer outside: the simulator used in acoustic design and the document-processing engine are available as standalone software systems.

Development areas

Speech and audio AI

Speech recognition, transcription and audio analysis in Hungarian and other languages; these models power the minutes feature of irat.ai and our audio technology products.

AI-based acoustic simulation

Machine-learning-accelerated finite-element and wave-based acoustic simulation for automotive and architectural applications; room-shape optimisation, parameter estimation, noise-propagation algorithms. This became Soundy.

Document and data processing

Large-scale processing of scanned and digital documents, structured data extraction, entity linking and relationship analysis — the processing engine of ÁTMR.

Legal language models and translation

Language models tuned to legal and public-administration text: answers to legal questions, Slovak–Hungarian translation that follows the field's terminology.

Data-sovereign AI infrastructure

Auditable, multi-model AI agent and orchestration platform that runs on your own infrastructure, for regulated environments.

Machine learning in engineering work

Our first AI application, in 2016, accelerated numerical computations; since then our own tools also support design, expert and technical-supervision work — analysing large volumes of documentation, frequency plans and network datasets.

Projects and publications

Thinking about a joint research or development project? Get in touch.

Contact