DNN-based performance measures for predicting error rates in automatic speech recognition and optimizing hearing aid parameters

verfasst von

A.M. Castro Martinez, Lukas Gerlach, Guillermo Payá-Vayá, Hynek Hermansky, Jasper Ooster, Bernd T. Meyer

Abstract

In several applications of machine listening, predicting how well an automatic speech recognition system will perform before the actual decoding enables the system to adapt to unseen acoustic characteristics dynamically. Feedback about speech quality, for instance, could allow modern hearing aids to select a speech source in complex acoustic scenes with the aim of enhancing the speech intelligibility of a target speaker. In this study, we look at different performance measures to estimate the word error rates of simulated behind-the-ear hearing aid signals and detect the azimuth angle of the target source in 180-degree spatial scenes. These measures derive from phoneme posterior probabilities produced by a deep neural network acoustic model. However, the more complex the model is, the more computationally expensive it becomes to obtain these measures; therefore, we assess how the model size affects prediction performance. Our findings suggest measures derived from smaller nets are suitable to predict error rates of more complex models reliably enough to be implemented in hearing aid hardware.

Organisationseinheit(en)

Fachgebiet Architekturen und Systeme

Externe Organisation(en)

Exzellenzcluster Hearing4all
Carl von Ossietzky Universität Oldenburg
Johns Hopkins University

Typ

Artikel

Journal

Speech communication

Band

106

Seiten

44-56

Anzahl der Seiten

13

ISSN

0167-6393

Publikationsdatum

01.2019

Publikationsstatus

Veröffentlicht

Peer-reviewed

Ja

ASJC Scopus Sachgebiete

Software, Modellierung und Simulation, Kommunikation, Sprache und Linguistik, Linguistik und Sprache, Maschinelles Sehen und Mustererkennung, Angewandte Informatik

Ziele für nachhaltige Entwicklung

SDG 3 – Gute Gesundheit und Wohlergehen

Elektronische Version(en)

https://doi.org/10.1016/j.specom.2018.11.006 (Zugang: Geschlossen)