Speech Dereverberation
Speech dereverberation aims to remove the unwanted echoes and reverberations from audio recordings, improving speech intelligibility and the performance of downstream tasks like speech recognition. Current research heavily utilizes deep learning, employing architectures like diffusion models, generative adversarial networks (GANs), and recurrent neural networks (RNNs), often incorporating multi-modal approaches that leverage visual information alongside audio. These advancements are significant for improving human-computer interaction, assistive technologies for hearing-impaired individuals, and the robustness of speech processing systems in real-world acoustic environments.
Papers
Neural Network-augmented Kalman Filtering for Robust Online Speech Dereverberation in Noisy Reverberant Environments
Jean-Marie Lemercier, Joachim Thiemann, Raphael Koning, Timo Gerkmann
A neural network-supported two-stage algorithm for lightweight dereverberation on hearing devices
Jean-Marie Lemercier, Joachim Thiemann, Raphael Koning, Timo Gerkmann
Customizable End-to-end Optimization of Online Neural Network-supported Dereverberation for Hearing Devices
Jean-Marie Lemercier, Joachim Thiemann, Raphael Koning, Timo Gerkmann