A DESIGN FOR ACTIVE NOISE CANCELLATION IN THE 2CC COUPLER Jiun-Hung Lin, Kuang-Jung Wu, Ping-Ting Liu, William

Chu, Shuenn-Tsong Young National Yang-Ming University, Institute of Biomedical Engineering No. 155, Sec. 2, Li-Nong St., Shih-Pai, Taipei(112), Taiwan, ROC E-mail: [email protected]
ABSTRACT Speech recognition in noise environment is always inefficient for hearing-impaired people who wear hearing aids. Many noise cancellations have been proposed and studied for such an issue. In this paper, we designed a prototype of hearing instrument coupler to imitate the equivalent volume of the human ear, and built a feedforward active noise cancellation system (FANCs) that had the algorithm implemented with a digital signal processor (DSP) controller. The hearing instrument coupler is used to study the FANCs efficiency applied to real hearing aids. It is a chamber with an effective volume of 2cc specified by International Electrotechnical Commission (IEC) standard. The FANCs includes a reference microphone, an error microphone, a canceling loudspeaker and a DSP controller. The results show that the hearing instrument coupler could simulate an ear with hearing aid, and the FANCs could suppress the unwanted noise effectively to enhance speech recognition. hearing aids are then equipped with certain noise suppression strategy [1] [6]. The most commonly employed strategies include the use of a directional microphone, amplitude compression circuitry, and multiband filters. The directional microphone has been shown as an effective method to improved the signal-to-noise ratio in hearing aids. It really enhances the speech intelligibility for the hearing-impaired people when the speech signals are happened in front of the microphone and the noise signals are happened in the rear. Over the past twenty years, lots of powerful digital signal processing methods have also been developed for reducing noises. Most of them are highly computationally intensive and are difficult to implement in small portable devices for real time processes until recently. However, with the increasing availability of inexpensive and high performance DSPs, it may be soon possible to implement digitalized noise cancellation in practical hearing aids. Adaptive noise cancellation [Widrow et al., 1975a] is a popular approach for noise reduction. Its basic concept is to process signals from two sensors and to reduce the undesired noise with adaptive filtering techniques. It always includes two signal inputs, an adaptive filter, and an error estimator for adaptive processes. Basing on the basic structure, we proposed another feedforward active noise cancellation system that will be applied to reduce the environmental noises in the 2cc coupler. The results will eventually be extended to apply in real hearing aids. 2. METHOD AND MATERIAL 2.1 The general concept of Active Noise Cancellation The general system of adaptive noise cancellation is illustrated in figure.1. The system receives two input signals from different microphones. The signal source S(n) consists of speech sound and noise. The noise source V(n) is a noise that is highly correlated with the noise in the signal source.The noise signal V(n) is shaped the adaptive filter and then subtracted from the signal source S(n) to produce the output of the noise canceller. The adaptive

1. INTRODUCTION The intelligibility of speech communication is dependent on the recognition of speech signals. For hearing-impaired people, the speech intelligibility may be heavily affected by many factors, including the intrinsic or extrinsic hearing impairment, hearing weakness, and various background noises. A hearing aid with appropriately prescription and properly fitting can significantly improve the ability of a hearing-impaired people to perceive speech in a noise-free and nonreverberant environment. The main function of general hearing aids is to modify the speech signals, and makes them more accessible and discriminable for the hearingimpaired people. However, such a benefit function can be easily exterminated by various background noises. In daily livelihood, many environmental noises have sufficient level to interfere with the hearing aids for accurate speech perception and recognition. Some

0-7803-7402-9/02/$17.00 2002 IEEE

IV - 3920

filter adjusts itself until the system output is with minimal error with S(n), which occurs when y is approximately the same as Vo.

Different applications of the ANCs, including real and simulated experiments, have been introduced. Many researchers have especially investigated the ANCs in ducts using DSP techniques [8] [10]. These ducts involved air-conditioning ducts, ventilation ducts, exhaust ducts, and so on. We would apply the some ANC techniques in a hearing instrument coupler to achieve the noise cancellation for hearing aids. The hearing instrument coupler is a device connecting a measurement and a receiver for hearing aids evaluation. It is specified as a chamber with an effective volume of 2cc to imitate the equivalent volume of the human ear. In the figure.3, it shows such a coupler with a cavity of 2cc.

Fig. 1. Basic concept of adaptive noise canceller.

The first acoustic ANCs was proposed by Lueg in a patent [1936] [7]. It employed a microphone to receive the original sound and an electronically driven loudspeaker to generate the canceling sound. The ANCs cancels the unwanted sound by generating an antisound with equal amplitude and opposite phase in an electroacoustic or electromechanical instrument. The unwanted sound and the antisound combine and result in the cancellation to each other. The physical concept of the active noise cancellation is illustrated in figure.2.

Fig. 3. The 3-D dimensions of a 2-cc coupler

The ANCs has different basic structure with feedforward control or feedback control. In this paper, we proposed and constructed a feedforward active noise cancellation system (FANCs) with the hearing instrument coupler to study noise cancellation in hearing aids. The FANCs structure with the hearing instrument coupler is illustrated in figure.4. A primary microphone close to the noise source senses a reference signal. The ANC controller uses the reference signal to generate a reverse signal with equal amplitude but opposite phase. The reverse signal then drives a canceling loudspeaker to generate the antisound that is used to cancel the unwanted noise within the coupler. The system is also equipped an error microphone to measure the residual noise signal. The residual noise signal is used to adjust the coefficients of the adaptive filter for optimizing the noise cancellation. Fig. 2. Physical concept of active noise cancellation.

2.2 The structure of FANCs in the 2cc coupler

IV - 3921

Fig. 4. The structure of FANCs with hearing instrument coupler

Fig. 6. The experimental prototype system in sound box.

3. PROTOTYPE DESCRIPTION The prototype system for acoustic noise cancellation was tested in the hearing instrument coupler with chamber dimension of 1.9mm 1.9mm 7.2mm. The side view of the coupler is shown on figure.5.

A loudspeaker was located in the sound proof box, and it was driven by an audio frequency generator (noise) from the hearing instrument analyzer. The primary microphone received the audio signals and drove the primary speaker to generate the reference signal via an amplifier. The canceling speaker then generated a reverse signal with equal amplitude but opposite phase to cancel the unwanted noise. The reverse signal was determined by the DSP controller in the coupler. Therefore, the coupler microphone measured the residual signals and delivered it to hearing instrument analyzer to investigate the sound canceling affections of the prototype system. 4. RESULTS FOR THE PROTOTYPE SYSTEM In the test experiment, we employed the pure signals, which sweep from 125Hz to 8kHz as the noise signals. The sampling frequency of DSP controller was set to 8kHz and its input signal was limited between -1.5v and 1.5v. The acoustic response with and without DSP controller is shown as figure.7. The maximal achieved cancellation was significantly about 18 dB for an input signal frequency of 180Hz on the 2cc coupler. 5. CONCLUSION To verify the active noise cancellation techniques on the small cavity of the hearing instrument coupler, we designed a prototype system and analyzed its residual signals in the sound proof box. The result showed that the prototype system achieved noise cancellation about 18 dB for a signal with frequency of 180Hz. There was also

Fig. 5. A side view of the 2cc coupler.

In order to avoid sound effect from free field, we tested the prototype system in a sound proof box. The measurement instrument used in the test was a hearing instrument analyzer manufactured with the IEC standard 118-7 by SIMENS corp. The whole test system was setup as figure.6.

IV - 3922

some noise elimination on other frequencies from 300Hz to 3kHz. However, the environmental noises are not only of aperiodic, but also they are of various spreads and with broadband frequencies. Such characteristics will heavily confuse our speech recognition. Furthermore, the ANC's efficiency in the small hearing instrument coupler may be affected by many other factors, including the nonlinear characteristics of the control microphone and canceling speaker, the sound absorptivity of the coupler wall, and the sound reverberation in the coupler. Simple FANCs might cause the system to be unstable. Therefore, we will propose the novel adaptive algorithms on the FANCs, and study its enhancement for speech recognition in the hearing aids.

[4] T.A. de Perez., Min Li, McAllister, H., and Black, N.D., Noise reduction and loudness compression in a wavelet modelling of the auditory system. IEEE International Caracas Conference on Devices, Circuits and Systems Proceeding., pp. 15-17, 2000. [5] Min Li, McAllister, H.G., Black, N.D., and De Perez, T.A., Perceptual time-frequency subtraction algorithm for noise reduction in hearing aids. IEEE International Conference on Biomedical Engineering., vol.48, pp. 979-988, 2001. [6] Trenas, M.A., Rutledge, J.C., and Whitmal, N.A., Waveletbased noise reduction and compression for hearing aids. BMES/EMBS Conference, 1999. Proceedings of the First Joint., vol.1, pp. 670, 1999. [7] P.Lueg, Process of silencing sound oscillations, U.S. Patent2 043 416, June 9, 1936. [8] S. M. Kuo and D. R. Morgan, Active Noise Control Systems-Algorithms and DSP Implementations. New York: Wi-ley, 1996. [9] S. M. Kuo and C. Chen, Implementation of adaptive filters with the TMS320C25 or the TMS320C30, in Digital Signal Processing Applications with the TMS320 Family, vol. 3, P. Papamichalis, Ed. Englewood Cliffs, NJ: Prentice-Hall, 1990, ch. 7, pp. 191-271. [10] S. J. Elliott and T. J. Sutton, Performance of feedforward and feedback systems for active control. IEEE Trans on Speech Audio Processing, vol. 4, pp. 214-223, 1996.

Fig. 7. The test result of the prototype system in the sound proof box. Solid line represents the operation when DSP controller turns off and dashed line represents the operation when DSP controller turns on.

6. REFERENCES
[1] Yanagisawa, K., Tanaka, Y., and Yamaura, I., Applying comb filter to noise reduction of hearing aid. IEEE International Conference on Systems, Man, and Cybernetics Proceedings- SMC '99., vol.6, pp. 352-357, 1999. [2] Dongwook Kim, and Inyoung Kim, Comparisons of speech enhancement algorithms for normal and sensorineural hearing impairment listeners. BMES/EMBS Conference, 1999. Proceedings of the First Joint, vol.1, pp. 655, 1999. [3] Yanagisawa, K., Tanaka, Y., and Yamaura, I., Detection of the fundamental frequency in noisy environment for speech enhancement of a hearing aid. IEEE International Conference on Control Applications Proceedings., vol.2, pp. 1330 - 1335, 1999.

IV - 3923