Chapter 5 Diagnostic Systems for Environmental Noise

5.1. Introduction
In order to measure orthogonal factors, SPL, Dt1, Tsub, IACC, tIACC, and WIACC [51-56], and also the running ACF of sound field at each seat in a scale model as well as in a real auditorium, a diagnostic system is developed. Based on the model of auditory-brain system which consists of the autocorrelation mechanism, the interaural crosscorrelation mechanism between the both auditory pathways, and the specialization of human cerebral hemispheres [51], a diagnostic system was developed. The system works on PC for Windows with an AD & DA converters, thus it is no need for special additional devices. After obtaining the binaural impulse responses, four orthogonal factors including the SPL, the initial time delay gap between the direct sound and the first reflection, the subsequent reverberation time and the IACC are analyzed. These factors are used for the calculation of both the scale values of global and individual subjective preferences. In addition to the four factors, two more factors, tIACC and WIACC, extracted from the interaural crosscorrelation function can be figured out for evaluating the image shift of sound source and the apparent source width [56], respectively. Also, the averaged sound energy for two ears, the effective duration, te, defined by the delay at which the envelope of normalized ACF becomes 0.1, and fine structures of autocorrelation function of sound signals including the magnitude of first maximum, f1, and its delay time, t1, of source signals are analyzed. Also, for the internet based measurement of environmental noise, this system may be utilized for identifying source signals and spatial information.

5.2. Outline of a diagnostic system
    Measurement of environmental noise is illustrated in Figure 5.1. The purpose is measuring an influence which an airport have on the region. This is an example of technical application of the concert hall measurement system [59, 60] for the internet technology.

Figure 5.1. The internet-oriented system for measuring environmental noise with dummy-head microphones; . The system identifies, for example, the aircraft noise.


    The measurement is performed by two channels on each measuring point, where the collected data are distinguished the aircraft noise from the others automatically [61, 62].
    In case of the aircraft noise, it is analyzed there, and is sent to central office through internet. Primary sensations, pitch, loudness and timbre, of a given source signal and sound field are described based on a model of auditory-brain system [51]. The model consists of both autocorrelation and interaural crosscorreration mechanisms. In order to describe timbre or quality of sound fields, for example, the human cerebral hemisphere specialization for the temporal and spatial factors is taken into consideration as the similar manner to the subjective preference.
    As this result, by sending only these parameters, it is not necessary to send whole sound data obtained by measurement. The diagnostic system may be applied to the internet by sampling environmental noise.

5.3. Identification of a noise source
    The noise is identified as that of an aircraft, an automobile, a factory and so on, by using four factors extracted from ACF; F(0), te, t1, and f1 (Figure 5.2, 5.3). It is the same method that is used for the speech intelligibility of single syllables [61]. Four factors, F(0), te, t1 and f1 are utilized for the identification. At our test, it has been found that the system can distinguish the engine sound between Porsche and Mercedes Diesel.

Figure 5.2. A practical example of determining effective duration of ACF defined by the ten-percentile delay, with the straight line-fitting envelope of ACF from 0 to -5 dB.


Figure 5.3. Definitions of the f1 and t1 for the autocorrelation function.

    A measurement is done automatically according to the condition a user sets. Measured data is analyzed in the background by MS Windows's multi-thread function, and each factor is extracted and the noise source is identified.
    By measuring an actual noise, this program can learn to get the template for identification of noise. It helps to raise the rate of correct identification. This learning function has a manual learning mode and an automatic learning mode. Matters that demand special attention is that a user must correct the noise source if the program makes wrong identification of it. If not, the program will do wrong learning and the probability of correct identification will be getting worse after that.
    The template of sound can be adjusted in "Noise Source Template" dialog. The average value, the maximum value and the minimum value about four factors, F(0), te, t1 and f1 are input. The maximum value and the minimum value are valid only when its check-box is checked. The measured data that is beyond the range of setting value will be removed from its template. The more the value in "Weighting" area is large, the more its factor has great influence in identification. (Appendix C)
    Measured data will be stored in the database, and it can be analyzed more detailer by "Acoustic Analyzing System". The data of "Acoustic Analyzing System" was shown as Figure 5.4. The data of Porsche's engine sound and Mercedes Diesel's one is difference clearly, having each characteristic. The data of template was shown as Figure 5.5.

Figure 5.4. Example of F(0), te, t1 and f1 of Porsche Turbo, Mercedes 200 diesel, and Ferrari analyzed by Acoustic Analyzing System.







Figure 5.5. Noise source template of Porsche Turbo and Mercedes 200 Diesel.



5.4. Further study of signal of environmental noise

There are four orthogonal factors of sound fields, namely, listening level LL, initial time delay gap between the direct sound and the first reflection Dt1, subsequent reverberation time Tsub, and IACC. The IACC (maximum of the IACF) is related to the subjective diffuseness. In addition to the IACC, tIACC (delay time at the IACC obtained) and WIACC (width of the maximum) extracted from the IACF are deeply related to the image shift of sound source and the apparent source width (ASW), respectively. All of these factors must be considered to describe the primary sensations as well as the subjective preference [62].

1. The noise level: listening level LL at both ears. (1)
LL = [Fll(0)Frr(0)]1/2 (1)

2. The spatial information, for example, may be expressed by
st = ft (LL, IACC, WIACC, tIACC)right (2)

| BACK | NEXT |