improving partition-block-based acoustic echo canceler in under-modeling scenarios

Abstract

Recently, a partitioned-block-based frequency-domain Kalman filter (PFKF) has been proposed for acoustic echo cancellation. Compared with the normal frequency-domain Kalman filter, the PFKF utilizes the partitioned-block structure, resulting in both fast convergence and low time-latency. We present an analysis of the steady-state behavior of the PFKF and found that it suffers from a biased steady-state solution when the filter is of deficient length. Accordingly, we propose an effective modification that has the benefit of the guaranteed optimal steady-state behavior. Simulations are conducted to validate the improved performance of the proposed method.

arXiv commentaccepted by interspeech2020

References in corpus (7)

• Adam: A Method for Stochastic Optimization
• Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
• Generative Adversarial Text to Image Synthesis
• Don't Decay the Learning Rate, Increase the Batch Size
• Cognitive Science in the era of Artificial Intelligence: A roadmap for reverse-engineering the infant language-learner
• Representations of language in a model of visually grounded speech signal
• Language learning using Speech to Image retrieval