IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 14, NO. 2, MARCH 2006 331 Active Noise Control System for Headphone Applications Sen M. Kuo, Senior Member, IEEE, Sohini Mitra, and Woon-Seng Gan, Senior Member, IEEE Abstract—This paper presents the design and implementation of an adaptive feedback active noise control (ANC) system for head- phone applications. The ideal position of the error microphone in the ear-cup was studied and determined experimentally, and music signals were used for adaptive system identification of the secondary path. The designed ANC headphone was implemented using the TMS320C32 digital signal processor for real-time ex- periments. Performance has been evaluated and compared with a high-end commercial ANC headphone using the same set of primary noises including real-world engine noises. Experiment results show the proposed ANC headphone achieves higher noise cancellation, especially for low-frequency harmonics. Index Terms—Active noise control (ANC), adaptive feedback ANC systems, ANC headphones, error sensor optimization, sec- ondary path modeling. I. INTRODUCTION W ITH the growth of technology and industry, acoustic noise problems have become more and more acute. Noise levels in human settings have come under scrutiny for reasons including health concerns and improvement of the quality of life. Common acoustic noises originate from the industrial products such as engines, blowers, fans, and trans- formers. For low-frequency noises, passive methods such as earmuffs are either ineffective or tend to be very expensive or bulky. Thus, the active noise control (ANC) systems [1]–[4], which efficiently attenuate low-frequency noises, have be- come an effective technique for designing ANC headphones to protect workers in noisy environments. The advent of fast computational devices such as digital signal processors has made the implementation of digital ANC headphone possible. ANC headphone cancels the primary noise by introducing a secondary “antinoise” of equal amplitude but opposite phase, thus attenuating undesired acoustic noises inside the ear-cups. In feedforward ANC headphones, a reference microphone is placed outside the ear-cup to pick up the primary noise. This ref- erence signal is then processed by the ANC system to gen- erate the control signal to drive a secondary loudspeaker in- side the ear-cup for producing antinoise. The error microphone monitors the performance of the ANC system by measuring the residual noise , which will be minimized by the adaptive al- gorithm. In this paper, we develop the adaptive feedback ANC (AFANC) system which uses the internal model (the estimated secondary path) to synthesize the reference signal instead Manuscript received January 5, 2003; revised September 14, 2005. Manu- script received in final form November 1, 2005. Recommended by Associate Editor R. Rajamani. S. M. Kuo and S. Mitra are with the Center for Acoustics and Vibration, Department of Electrical Engineering, Northern Illinois University, DeKalb, IL 60115 USA (e-mail: kuo@ceet.niu.edu). W.-S. Gan is with the School of Electrical and Electronics Engineering, Nanyang Technological University, Singapore 639798. Digital Object Identifier 10.1109/TCST.2005.863667 Fig. 1. Block diagram of adaptive feedback ANC system. of using a reference microphone. Thus, the AFANC system uses only one error microphone per ear-cup, and results in a low-cost and compact ANC headphone design [5]. A comparative study of various AFANC systems using four different adaptive algo- rithms for headset applications is reported in [6]. In practical ANC systems, the secondary path (see Fig. 1) includes the digital-to-analog converter (DAC), recon- struction filter, power amplifier, loudspeaker, acoustic path from loudspeaker to error microphone, error microphone, preamplifier, antialiasing filter, and analog-to-digital converter (ADC). The filtered-X least-mean-square (FXLMS) algorithm [7] places the secondary-path estimate in the reference signal path to the weight update of the algorithm. This secondary-path model is usually estimated by adaptive system identification using white noise as an excitation signal, which is annoying for headphone applications. The robustness of AFANC headset can be improved by adding an analog feedback loop to handle large secondary path flocculation [8]. In this paper, we study the feasibility of using music as an excitation signal for modeling secondary path and show it can achieve similar results as white noise and chirp signal. The performance of ANC systems is determined by the mag- nitude response of the secondary path, which can be affected by the placement of the error microphone. In this paper, the op- timum location of the error microphone is subject to the flat magnitude response of the secondary path, is found by exper- iments, and explained by the spectral filtering of pinna. An ANC headphone was built for experiments by inserting an error microphone inside each ear-cup of a commercial audio headphone with a built-in loudspeaker. The AFANC algorithm was coded in assembly program and implemented on the Spec- trum Signal Processing Inc. TMS320C32 processor board with 16-bit ADC and DAC on the 4 Channel Analog I/O Board for analog interface. Each of the four input channels and two output channels are provided with third-order Butterworth anti-aliasing filters. The cutoff frequency of these filters is set at 3.38 kHz with filter parameters given in the user’s manual. The designed headphone was tested using sinusoidal noises generated by a 1063-6536/$20.00 © 2006 IEEE