The 2001 International Congress and Exhibition on Noise Control Engineering The Hague, The Netherlands, 2001 August 27-30 In-Ear Microphone Equalization Exploiting an Active Noise Control.
Nils Westerlund, Mattias Dahl, Ingvar Claesson Blekinge Institute of Technology Department of Telecommunications and Signal Processing Soft Center S-372 25 RONNEBY
Abstract
A pair of ear-muffs that employs active noise control (ANC) for noise reduction, substan- tially reduces the influence of the low frequencies inside the cap. This implies an indirect high-pass filtering of the sound in the external auditory canal (EAC). This paper shows that the above mentioned high-pass filtering property is convenient when combining an ANC headset with an in-ear microphone (ear-mic) for communication purposes since the speech signal inside the EAC is a low-pass filtered version of the speech signal at the mouth. Hence, the in-ear speech signal is to some extent restored by the ANC high-pass filtering, the quality of the speech signal in the auditory canal is improved and the speech intelligibility is increased. The equalization will also decrease the demand of dynamic range and resolution of electronics used. By that, combining an active ear-muff with an ear-mic serves two purposes: Protecting the user from harmful noise and enables the user to communicate over some channel using the speech signal in the auditory canal.
1. Introduction
Many occupations of today requires the usage of personal preservative equipment such as a pair of ear-muffs to damp high sound pressure levels. At the same time, there is often a need to be able to communicate via some communication equipment. The primary concern with noise is not only the potential risk of damage to the hearing. Noise with a high sound pressure level is also annoying and during periods of long exposure, it causes fatigue, vertigo, nausea and loss of concentration. Another effect caused by low frequency noise is the masking effect it has on speech. This masking can significantly decrease speech intelligibility which in turn forces the user to increase the volume in the communication system. This alone can lead to sound pressure levels that are damaging to the human ear [1]. Active noise control is an attractive alternative to passive reduction of unwanted noise since the latter implies heavy and bulky absorbers. ANC is an effective way of cancelling noise at frequencies below approximately 1000 Hz [2]. Today, ear-muffs equipped with ANC are commercially available but there are still problems when trying to communicate in a noisy environment. The idea that a micro- phone for communication purposes can be placed at other locations of the body than just in front of the mouth, is by no means a new one. Different locations of the body, such as the throat or forehead, has been used to attach an accelerometer (i.e. a tactile microphone) [3], [4]. The external auditory canal (EAC) has also been investigated as an alternative microphone location [5]-[9]. However, the combination of an ANC-equipped pair of ear-muffs and a microphone placed in the EAC is not as thoroughly investigated. An audiometric earphone system which employs ANC implemented in a foam plug is introduced in [10] but this solution is not intended for communication purposes. In this paper, the advantages of combining a pair of ANC-equipped ear-muffs with a microphone for communication purposes placed inside the EAC (ear-mic) is investigated.
2. Presentation of the Problem
The combination of ANC and ear-mic has some advantages since the mouth-to-ear chan- nel represents a fairly simple low-pass filtering. Hence, the mouth-to-ear channel is to some extent equalized by the ANC since the noise control represents a high-pass filter- ing. Due to this, the quality of the speech signal in the EAC is improved and the speech intelligibility is increased. Furthermore, the demands on dynamic range and resolution of the electronics used, will decrease. Note also that this indirect high-pass filtering does not introduce any additional delay on the speech signal. This is advantageous since this equipment is intended to be used in a mobile communication system. Since the channel equalizer is designed to operate in such a large system, it is desirable to reduce the delay caused by the filtering and in this way minimize the total delay introduced by the whole system. Another aspect of the problem is that the speech signal inside the EAC, is dominated by low frequency components and the high frequency components are heavily damped. This raises the demands on the dynamic range of the equipment used. Some sort of inverse filtering or channel equalization may be desirable to amplify higher frequencies.
3. Data Acquisition and Method
Narrow canals were drilled in a pair of custom moulded, acrylic ear-plugs and microphone probes were inserted into these canals. The effects of using microphone probes are thoroughly investigated in [11]. Two small-size microphones (Sennheiser) were attached to the other end of the probes, See Figure 1. Figure 1: Custom moulded acrylic ear-plug with microphone and microphone probe attached.
Two different ear-muffs were evaluated: A pair of modified Hellberg ear-muffs and Bose QC-1 ear-muffs. Both the Hellberg and the Bose ear-muffs were equipped with a switch to enable/disable the ANC. Three different noise environments were used: No noise (speech only), noise from the outside of a helicopter and noise from a helicopter cockpit. Both of these types of noise are normally dominated by tonal components in the lower frequency range (< 500 Hz) [12]. A loudspeaker was used to reproduce the noisy environment. Figure 2 illustrates the measurement setup and signal paths. The analog speech signals from the mouth and the EAC, xM (t) and xE(t) respectively, were recorded to a DAT recorder. The resulting digital signals xM (n) and xE(n) were transferred to a Matlab-file and Welch’s method were used to calculate the power spectral densities (PSD) of the signals, PM (f) and PE(f).
5 4 1 2 3
Analog Feedback ANC
DAT Recorder
xE(t) xE(n) PE(f) ADC PSD x (n) xM(t) M 6 PM(f)
Figure 2: Measurement setup and signal paths. (1) Ear-mic. (2) Custom moulded acrylic ear-plug. (3) ANC reference microphone and loudspeaker. (4) Ear-muff. (5) Surrounding noise. (6) Reference microphone at mouth.
The basic idea is that a spectral weighting function (i.e. a gain function) G(f) could be calculated using the expression
PM (f) G(f) = (1) PE(f) The function G(f) could then be used to shape the PSD of the speech signal inside the EAC to a PSD more like the one at the mouth and in that way improve the quality of the speech.
4. Results
Figure 3 and Figure 4 shows the PSD:s for the sound inside the EAC. It is clear that the low frequency components of the signal inside the EAC are damped to some extent when ANC is enabled. This low frequency damping makes the speech signal more intelligible.
100
50 Magnitude [dB]
0 0 50 100 150 200 250 300 350 400 450 500 Frequency [Hz] 100
50 Magnitude [dB]
0 0 50 100 150 200 250 300 350 400 450 500 Frequency [Hz] 100
50 Magnitude [dB]
0 0 50 100 150 200 250 300 350 400 450 500 Frequency [Hz]
Figure 3: Power spectral densities of speech signal inside the EAC using Hellberg ear-muffs. Solid line: ANC disabled. Dashed line: ANC enabled. (Upper plot) Speech with no surrounding noise. (Mid plot) Speech and noise from the outside of a helicopter. (Lower plot) Speech and noise from helicopter cockpit.
To further improve the speech signal quality and intelligibility, gain functions were calculated according to (1). The gain function with ANC enabled and disabled for both Hellberg and Bose ear-muffs, are plotted in Figure 5.
Conclusions
Since the head represents a fairly simple low-pass system and the active noise control system damps low frequencies, the speech picked up inside the EAC becomes more intel- ligible when the active noise control is enabled. The use of a “gain function” calculated from the PSD:s of the input and output signal can also improve the speech quality. 100
50 Magnitude [dB]
0 0 50 100 150 200 250 300 350 400 450 500 Frequency [Hz] 100
50 Magnitude [dB]
0 0 50 100 150 200 250 300 350 400 450 500 Frequency [Hz] 100
50 Magnitude [dB]
0 0 50 100 150 200 250 300 350 400 450 500 Frequency [Hz]
Figure 4: Power spectral densities of speech signal inside the EAC using Bose ear-muffs. Solid line: ANC disabled. Dashed line: ANC enabled. (Upper plot) Speech with no surrounding noise. (Mid plot) Speech and noise from the outside of a helicopter. (Lower plot) Speech and noise from helicopter cockpit.
These two factors can form a basis for further studies in the ear-mic area and facilitate as well as improve other methods used to equalize the mouth-to-ear channel. For more information, visit http://www.bth.se/its.
References
1. S. Johansson, “Active Control of Propeller Induced Noise in Aircraft”,Doctoral Dissertation, Blekinge Institute of Technology, Dept. of Telecommunications and Signal Processing, 2000.
2. M. Winberg, S. Johansson, T. Lag¨o, I. Claesson, “A New Passive/Active Hybrid Headset for a Helicopter Application”, International Journal of Acoustics and Vi- bration, 1999, no. 2, pp. 51-58.
3. D. W. Martin, “Magnetic Throat Microphone of High Sensitivity”, Journal of the Acoustical Society of America, 1947, no. 1, pp. 43-50.
4. H. M. Moser, H. J. Oyer,“Relative Intesities of Sounds at Various Anatomical Lo- cations of the Head and Neck during Phonation of the Vowels”, Journal of the Acoustical Society of America, 1958, no. 4, pp. 275-277.
5. R. D. Black, “Ear-Insert Microphone”, Journal of the Acoustical Society of America, 1957, no. 2, pp. 260-264. 30
20
10
0 Magnitude [dB] −10
−20 0 500 1000 1500 2000 2500 3000 3500 Frequency [Hz]
30
20
10
0 Magnitude [dB] −10
−20 0 500 1000 1500 2000 2500 3000 3500 Frequency [Hz]
Figure 5: (Upper plot) Gain function for Hellberg ear-muffs. Solid line: ANC disabled. Dashed line: ANC enabled. (Lower plot) Gain function for Bose ear-muffs. Solid line: ANC disabled. Dashed line: ANC enabled.
6. H. J. Oyer, “Relative Intelligibility of Speech Recorded Simultaneously at the Ear and Mouth”, Journal of the Acoustical Society of America, 1955, no. 6, pp. 1207- 1212.
7. H. Ono, “Improvement and evaluation of the vibration pick-up-type ear microphone and two-way communication device”, Journal of the Acoustical Society of America, September 1977, no. 3, pp. 760-768.
8. S. Aoki, K. Mitsuhashi, Y. Nishino, T. Sakurai, “Noise-suppressing Compact Mi- crophone/Receiver Unit”, NTT-Review, 1998, no. 6, pp. 102-108.
9. S. Aoki, K. Mitsuhashi, Y. Nishino, “Super-compact microphone/receiver unit for noisy environments”, Journal of the Acoustical Society of Japan, 1999, no. 5, pp. 381-383.
10. B. Rafaely, M. Furst, “Noise-suppressing Compact Microphone/Receiver Unit”, IEEE Transactions on Speech and Audio Processing, 1996, no. 3, pp. 224-230.
11. P. A. Hellstr¨om, “Miniature microphone probe tube measurements in the external auditory canal”, Journal of the Acoustical Society of America, 1993, no. 2, pp. 907-919.
12. T. L. Lag¨o, “Frequency Analysis of Helicopter Sound in the AS323 Super Puma”, Technical Report, ISSN 1103-1581, ISRN HKR-RES-96/8-SE, 1996.