BEA-Base: A Benchmark for ASR of Spontaneous Hungarian

no code implementations • LREC 2022 • Peter Mihajlik, Andras Balog, Tekla Etelka Graczi, Anna Kohari, Balázs Tarján, Katalin Mady

Hungarian is spoken by 15 million people, still, easily accessible Automatic Speech Recognition (ASR) benchmark datasets – especially for spontaneous speech – have been practically unavailable.

Automatic Speech Recognition Automatic Speech Recognition (ASR) +3

Deep Transformer based Data Augmentation with Subword Units for Morphologically Rich Online ASR

no code implementations • 14 Jul 2020 • Balázs Tarján, György Szaszák, Tibor Fegyó, Péter Mihajlik

Recently Deep Transformer models have proven to be particularly powerful in language modeling tasks for ASR.

Language Modelling Text Augmentation +1

On the Effectiveness of Neural Text Generation based Data Augmentation for Recognition of Morphologically Rich Speech

no code implementations • 9 Jun 2020 • Balázs Tarján, György Szaszák, Tibor Fegyó, Péter Mihajlik

In our recent work we have significantly improved the online performance of a conversational speech transcription system by transferring knowledge from a Recurrent Neural Network Language Model (RNNLM) to the single pass BNLM with text generation based data augmentation.

Automatic Speech Recognition Automatic Speech Recognition (ASR) +4

Investigation on N-gram Approximated RNNLMs for Recognition of Morphologically Rich Speech

no code implementations • 15 Jul 2019 • Balázs Tarján, György Szaszák, Tibor Fegyó, Péter Mihajlik

Recognition of Hungarian conversational telephone speech is challenging due to the informal style and morphological richness of the language.

Language Modelling MORPH +2