Výsledky vyhledávání - "Fegyó, Tibor"

Report

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

Autor: Tarján, Balázs, Szaszák, György, Fegyó, Tibor, Mihajlik, Péter

Recently Deep Transformer models have proven to be particularly powerful in language modeling tasks for ASR. Their high complexity, however, makes them very difficult to apply in the first (single) pass of an online system. Recent studies showed that

Externí odkaz: http://arxiv.org/abs/2007.06949

Zobrazit plný text záznamu

Report

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

Autor: Tarján, Balázs, Szaszák, György, Fegyó, Tibor, Mihajlik, Péter

Advanced neural network models have penetrated Automatic Speech Recognition (ASR) in recent years, however, in language modeling many systems still rely on traditional Back-off N-gram Language Models (BNLM) partly or entirely. The reason for this are

Externí odkaz: http://arxiv.org/abs/2006.05129

Zobrazit plný text záznamu

Report

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

Autor: Tarján, Balázs, Szaszák, György, Fegyó, Tibor, Mihajlik, Péter

Recognition of Hungarian conversational telephone speech is challenging due to the informal style and morphological richness of the language. Recurrent Neural Network Language Model (RNNLM) can provide remedy for the high perplexity of the task; howe

Externí odkaz: http://arxiv.org/abs/1907.06407

Zobrazit plný text záznamu