Retrieving tract variables from acoustics: A comparison of different machine learning strategies

Vikramjit Mitra, Hosung Nam, Carol Y. Espy-Wilson, Elliot Saltzman, Louis Goldstein

Research output: Contribution to journalArticle

37 Citations (Scopus)

Abstract

Many different studies have claimed that articulatory information can be used to improve the performance of automatic speech recognition systems. Unfortunately, such articulatory information is not readily available in typical speaker-listener situations. Consequently, such information has to be estimated from the acoustic signal in a process which is usually termed speech-inversion. This study aims to propose and compare various machine learning strategies for speech inversion: Trajectory mixture density networks (TMDNs), feedforward artificial neural networks (FF-ANN), support vector regression (SVR), autoregressive artificial neural network (AR-ANN), and distal supervised learning (DSL). Further, using a database generated by the Haskins Laboratories speech production model, we test the claim that information regarding constrictions produced by the distinct organs of the vocal tract (vocal tract variables) is superior to flesh-point information (articulatory pellet trajectories) for the inversion process.

Original languageEnglish
Article number5570879
Pages (from-to)1027-1045
Number of pages19
JournalIEEE Journal on Selected Topics in Signal Processing
Volume4
Issue number6
DOIs
Publication statusPublished - 2010 Dec 1
Externally publishedYes

    Fingerprint

Keywords

  • Articulatory phonology
  • articulatory speech recognition (ASR)
  • artificial neural networks (ANNs)
  • coarticulation
  • distal supervised learning
  • mixture density networks
  • speech inversion
  • task dynamic and applications model
  • vocal-tract variables

ASJC Scopus subject areas

  • Signal Processing
  • Electrical and Electronic Engineering

Cite this