Speech reconstruction using a deep partially supervised neural network

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-based systems have been hampered by the limited amount of training data available from individual voice-loss patients.

We propose a novel deep neural network structure that allows a partially supervised training approach on spectral features from smaller datasets, yielding very good results compared to the current state-of-the-art.