Helium Speech Correction Algorithm Based on Deep Neural Networks
Dongmei Li, Shibing Zhang, Lili Guo, Yonghong Chen · 2020
Unscrambling helium speech is an important technology in saturation diving. This paper focused on the unscrambling technology of helium speech and proposed a helium speech correction algorithm that is based on deep neural networks. It analyzed the features of helium speech and extracted the four features, formant frequency, formant bandwidth, formant amplitude and pitch, from the helium speech as the characteristic inputs of the deep neural networks. The back propagation is used to train the neural network in the correction algorithm. Simulation results show that the algorithm proposed would correct the distortions of helium speech effectively.