Voice Activity Detection for Ultrasound-based Silent Speech Interfaces using Convolutional Neural Networks

TL;DR

This paper develops a CNN-based voice activity detection method for ultrasound-based silent speech interfaces, improving speech/silence classification accuracy and analyzing the impact of silence in training data.

Contribution

It introduces a CNN classifier for ultrasound tongue images to distinguish speech from silence, enhancing silent speech interface performance.

Findings

Achieved approximately 85% classification accuracy.

Attained around 86% AUC score.

Silence in training data affects speech reconstruction quality.

Abstract

Voice Activity Detection (VAD) is not easy task when the input audio signal is noisy, and it is even more complicated when the input is not even an audio recording. This is the case with Silent Speech Interfaces (SSI) where we record the movement of the articulatory organs during speech, and we aim to reconstruct the speech signal from this recording. Our SSI system synthesizes speech from ultrasonic videos of the tongue movement, and the quality of the resulting speech signals are evaluated by metrics such as the mean squared error loss function of the underlying neural network and the Mel-Cepstral Distortion (MCD) of the reconstructed speech compared to the original. Here, we first demonstrate that the amount of silence in the training data can have an influence both on the MCD evaluation metric and on the performance of the neural network model. Then, we train a convolutional neural network classifier to separate silent and speech-containing ultrasound tongue images, using a conventional VAD algorithm to create the training labels from the corresponding speech signal. In the experiments our ultrasound-based speech/silence separator achieved a classification accuracy of about 85\% and an AUC score around 86\%.