Online reverberation time and clarity estimation in dynamic acoustic conditions

Abstract

Previously proposed methods for estimating acoustic parameters from reverberant, noisy speech signals exhibit insufficient performance under changing acoustic conditions. A data-centric approach is proposed to overcome the limiting assumption of fixed source–receiver transmission paths. The obtained solution significantly enlarges the scope of potential applications for such estimators. The joint estimation of reverberation time RT60 and clarity index C50 in multiple frequency bands is studied with a focus on dynamic acoustic environments. Three different convolutional recurrent neural network architectures are considered to solve the tasks of single-band, multi-band, and multi-task parameter estimation. A comprehensive performance evaluation is provided that highlights the benefits of the proposed approach.

On the Gains and Losses of Multimedia-Assisted Instruction Technology in College Music Teaching Practice

“…The BN is applied to the music teaching model in this paper without worrying about the Dead ReLU problem in ReLU when a high learning rate is set. It can also use a lower dropout to improve the training speed and is not sensitive to weight initialization (Götz et al, 2023). Therefore, the author's model uses Xavier weights as the initialization method (Useche & Hurtado, 2019).…”

Spatial Subtraction of Reflections from Room Impulse Responses Measured with a Spherical Microphone Array

“…Blind parameter estimation has been a topic of research for many years; a comparison of different methods was made through the ACE challenge [18]. Since then, several machine-learning-based algorithms for parameter estimation have been proposed, such as [19], [20]. So far, the motivation for parameter estimation was not using the parameters to render sound but to inform other algorithms, for example for speech enhancement or recognition, in order to improve their performance.…”