Blind Reverberation Time Estimation Using a Convolutional Neural Network
Hannes Gamper, Ivan Tashev · 2018
The reverberation time of an acoustic environment is a useful parameter for applications including source localisation, speech recognition and mixed reality. However, estimating the reverberation time blindly and on the fly remains a challenge. Here we propose formulating the estimation as a regression problem and using a convolutional neural network (CNN) to estimate the reverberation time directly from a four second long single-channel recording of reverberant speech in noise. Evaluation on the ACE Challenge data corpus suggests that the proposed method is computationally efficient and outperforms state-of-the-art methods.