ViVoVAD: a Voice Activity Detection Tool based on Recurrent Neural Networks

Autores/as

  • Pablo Gimeno Jordán Universidad de Zaragoza
  • Ignacio Viñals Bailo
  • Alfonso Ortega Giménez
  • Antonio Miguel Artiaga
  • Eduardo Lleida Solano

DOI:

https://doi.org/10.26754/jji-i3a.003524

Resumen

Voice Activity Detection (VAD) aims to distinguish
correctly those audio segments containing human
speech. In this paper we present our latest approach
to the VAD task that relies on the modelling
capabilities of Bidirectional Long Short Term
Memory (BLSTM) layers to classify every frame in
an audio signal as speech or non-speech

Descargas

Publicado

2019-05-20

Número

Sección

Artículos (Tecnologías de la Información y las Comunicaciones)