Transformer-based Acoustic Modeling for Hybrid Speech Recognition

We propose and evaluate transformer-based acoustic models (AMs) for hybrid speech recognition. Several modeling choices are discussed in this work, including various positional embedding methods and an iterated loss to enable training deep transformers. We also present a preliminary study of using l...

Ausführliche Beschreibung

Bibliographische Detailangaben

Veröffentlicht in:	arXiv.org 2020-04
Hauptverfasser:	Wang, Yongqiang, Abdelrahman, Mohamed, Le, Duc, Liu, Chunxi, Xiao, Alex, Mahadeokar, Jay, Huang, Hongzhao, Tjandra, Andros, Zhang, Xiaohui, Zhang, Frank, Fuegen, Christian, Zweig, Geoffrey, Seltzer, Michael L
Format:	Artikel
Sprache:	eng
Schlagworte:	Computer Science - Computation and Language Modelling Neural networks Speech recognition Transformers Voice recognition
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!