Blind Reverberation Time Estimation in Dynamic Acoustic Conditions
This work addresses the challenge of blind reverberation time estimation for applications in real-world scenarios where acoustic conditions are not static, representing an incremental improvement over prior methods.
The paper tackled the problem of estimating reverberation time in dynamically changing acoustic environments, showing that existing deep neural network methods perform poorly and proposing a novel training data generation approach that significantly improves the ability to track temporal changes.
The estimation of reverberation time from real-world signals plays a central role in a wide range of applications. In many scenarios, acoustic conditions change over time which in turn requires the estimate to be updated continuously. Previously proposed methods involving deep neural networks were mostly designed and tested under the assumption of static acoustic conditions. In this work, we show that these approaches can perform poorly in dynamically evolving acoustic environments. Motivated by a recent trend towards data-centric approaches in machine learning, we propose a novel way of generating training data and demonstrate, using an existing deep neural network architecture, the considerable improvement in the ability to follow temporal changes in reverberation time.