AI transcription In-house product
Multilingual AI meeting assistant
From meetings and online calls, in 50+ languages, with our own three-level processing, continuous retraining and secure data handling.
The project
Our own product, originally part of SRT. The system has to handle, all at once, multilingual meetings, technical terms and acronyms, and the protection of internal information, so that recordings turn into usable documentation, not just raw transcripts.
What we built
01Multilingual transcripts: in 50+ languages, including Hungarian, from meetings or any audio recording
02Our own three-level processing: understands Hungarian text too, and learns exceptions, acronyms and domain-specific terms
03Continuous, automatic retraining: learns from corrections on its own, with a success rate above 90%
04Automatic summaries: summaries, functional specifications and documentation from the transcript
05Own server, EU data centre or cloud: when run on your own server, data stays within the company; it can also be deployed in an EU data centre, or run in the cloud under licence
06Data protection: operation complies with the GDPR and data protection rules
Domain expertise gained
Multilingual speech recognition from recording, through Hungarian exceptions and technical terms, to finished documentation.
50+ languages Hungarian-language processing Domain-specific vocabulary learning Continuous retraining Own server, EU data centre or cloud GDPR-compliant data handling
Where it can be used
IN THIS PROJECT Our own meetings
Recording: client and internal meetings Becomes: summary, specification
Healthcare
Recording: consultations Becomes: documentation
Law firms
Recording: client meetings Becomes: notes
HR
Recording: interviews Becomes: summary, assessment
Research
Recording: focus groups, interviews Becomes: summary
International companies
Recording: multilingual meetings Becomes: a summary in the language you choose
