Unsupervised Training on a Large Amount of Arabic Broadcast News Data

The unsupervised training we carried out on the 1,858-hour untranscribed Arabic broadcast news (BN) data yields a sizable gain. However, this gain is only about half of that achieved on the 1,900-hour English BN data. This paper presents our efforts that aim at enlarging the gain on the Arabic data....

Ausführliche Beschreibung

Bibliographische Detailangaben

Hauptverfasser:	Ma, J., Matsoukas, S.
Format:	Tagungsbericht
Sprache:	eng
Schlagworte:	Arabic broadcast news Broadcast technology Broadcasting confidence estimation Design methodology Error analysis Inspection Mutual information Natural languages Neural networks Radio access networks Speech speech recognition unsupervised training
Online-Zugang:	Volltext bestellen
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!