Global SNR Estimation of Speech Signals for Unknown Noise Conditions Using Noise Adapted Non-Linear Regression

The performance of speech technologies deteriorates in the presence of noise. Additionally, we need these technologies to be able to operate across a variety of noise levels and conditions. SNR estimation can guide the design and operation of such technologies or can be used as a pre-processing tool in database creation (e.g. identify/discard noisy signals). We propose a new method to estimate the global SNR of a speech signal when prior information about the noise that corrupts the signal, and speech boundaries within the signal, are not available. To achieve this goal, we train a neural network that performs non-linear regression to estimate the SNR. We use energy ratios as features, as well as ivectors to provide information about the noise that corrupts the signal. We compare our method against others in the literature, using the Mean Absolute Error (MAE) metric, and show that our method outperforms them consistently.