학술논문
A Parallel Analog and Digital Adaptive Feedforward Controller for Active Noise Control
Document Type
Periodical
Author
Source
IEEE/ACM Transactions on Audio, Speech, and Language Processing IEEE/ACM Trans. Audio Speech Lang. Process. Audio, Speech, and Language Processing, IEEE/ACM Transactions on. 32:1100-1108 2024
Subject
Language
ISSN
2329-9290
2329-9304
2329-9304
Abstract
Digital adaptive controllers are widely used for feedforward active noise control, especially in headphones. In such applications, the secondary path delay, including the sampling and reconstruction effects, must be shorter than the primary path delay to maintain good broadband performance. A mixed analog and digital adaptive feedforward controller is developed to eliminate the added delay of the sampling and reconstruction. The analog controller is based on a state-filtered adaptive linear combiner, while the digital one uses an adaptive finite-impulse-response filter. It is shown that both filters can be adapted using the normalized filtered-reference LMS algorithm but with different secondary path models. A method to design the analog state-filter based on Padé’s approximation is described. The performance of the proposed controller with two analog states, the direct feedthrough and a 0.3 milliseconds delay, is assessed and compared to the separate analog or digital controllers in a controlled environment. The results highlight that adding the analog delay improves the digital controller performance by about 5 dB in this application, regardless of the primary noise direction.