RT Journal Article
A1 Ian McLoughlin
A1 Jingjie Li
A1 Yan Song
A1 Hamid R. Sharifzadeh

PB iet
T1 Speech reconstruction using a deep partially supervised neural network
JN Healthcare Technology Letters
VO 4
IS 4
SP 129
OP 133
AB Statistical speech reconstruction for larynx-related dysphonia has achieved good performance using Gaussian mixture models and, more recently, restricted Boltzmann machine arrays; however, deep neural network (DNN)-based systems have been hampered by the limited amount of training data available from individual voice-loss patients. The authors propose a novel DNN structure that allows a partially supervised training approach on spectral features from smaller data sets, yielding very good results compared with the current state-of-the-art.
K1 larynx related dysphonia
K1 statistical speech reconstruction
K1 DNN structure
K1 deep partially supervised neural network
K1 partially supervised training approach
K1 restricted Boltzmann machine arrays
K1 Gaussian mixture models
K1 voice-loss patients
DO https://doi.org/10.1049/htl.2016.0103
UL https://digital-library.theiet.org/;jsessionid=4dktuiphhho36.x-iet-live-01content/journals/10.1049/htl.2016.0103
LA English
SN
YR 2017
OL EN