MITLL 2015 Language Recognition Evaluation System Description

Abstract

The training data segments for the mandatory fixed training task made available by NIST was derived primarily from previously released data sources (Callhome, Callfriend, Mixer3, Switchboard, and VOA). Error! Reference source not found. shows a breakdown of the numbers of cuts and the speech duration (post-SAD) available for each language in the fixed training data set.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 27, 2016
Accession Number
AD1034036

Entities

People

  • Doug Sturim
  • Douglas Reynolds
  • Elizabeth Godoy
  • Elliot Singer
  • Frederick Richardson
  • Najim Dehak
  • Pedro Torres-carrasquillo
  • Stephen Shum

Organizations

  • MIT Lincoln Laboratory

Tags

Communities of Interest

  • Energy and Power Technologies
  • Human Systems

DTIC Thesaurus Topics

  • Air Force
  • Artificial Intelligence Software
  • Automated Speech Recognition
  • Bayesian Networks
  • Computer Programs
  • Data Sets
  • Department Of Defense
  • Dimensionality Reduction
  • Information Science
  • Language
  • Machine Learning
  • Neural Networks
  • Recognition
  • Statistics
  • Test Sets
  • United States
  • United States Government

Readers

  • Business Analytics
  • Speech Processing/Speech Recognition.