A Novel Method for Objective Evaluation of Interior Sound in a Passenger Car and Its Application to the Design of Interior Sound in a Luxury Passenger Car.
In previous research, the sound quality of interior sound has been presented with powerful and pleasant for the existing passenger car . It is practically impossible to make the interior sound powerful and pleasant with the traditional design method based on the structural design of a car body because there are many limitations such as weighting and fuel efficiency. Recently ASD technology has been employed to design the interior sound because it is easy to design and to generate the interior sound that the customers prefer with just audio sound system . However, the sound metric to express the powerfulness and pleasantness is necessary for the design of the sound with these sensations. The previous studies show that the booming sensation and the rumbling sensation are correlated to the sound quality of interior sound [3, 4] and the ratio of spectrum decay is correlated to the pleasantness of the horn sound and the music sound [5, 6]. The spectrum decay means the decay ratio of the sound pressure level per octave frequency. In automotive application, the decay ratio of sound pressure level per integer order can be used similarly. The booming sound is related to the sound pressure level of the interior sound and its frequency. Rumbling is related to the modulation of the interior sound. Therefore, in the paper, not only the booming index (BI) and modulation index (MI) are developed as the sound metrics of powerfulness and applied to the development of powerful index but also the ratio of order decay based on the spectrum decay is developed as the sound metric of pleasantness and applied to the development of pleasant index of a vehicle. Based on these results, systematic way to enhance sound character of interior engine sound will be introduced in the companion paper .
During acceleration and deceleration of a vehicle, a powerful tonal sound with low frequency less than 250 Hz occurs inside of cabin. This sound is called "booming sound" . The booming sound is a vibro-acoustic sound caused by the resonance between the structure-vibration modes and cavity modes of a car. The sensation of the booming sound is perceived as a dynamic sound or powerful sound because it compresses the passenger's ear with high sound pressure level. However, if the sound pressure level is too high, the interior sound is perceived as an annoyance car while the interior sound of a car without booming sensation is perceived as a quiet car. Therefore, the booming sound with proper sound pressure level is required to obtain the powerful sensation or dynamic sensation . The perception of tonal sound such as a booming sound is changed depending on the frequency. If the frequencies of two tonal sounds with the same sound pressure level are different each other, the perception for these sounds are different. Fig. 1 shows the pitch effect. All tone signals shown in Fig. 1 have the same sound pressure level of 80dB but the frequencies are different. Despite all signals have the same sound pressure level, people feel differently depending on the frequency. This different perception is due to the pitch effect of a tonal sound [7, 8]. Therefore, there are two major factors for the control of the perception for booming sound. One is the deference between the sound pressure level of the order component associated with a booming sound and that of its side orders, the other one is the frequency of the booming sound.
According to the Zwicker's empirical data , the relative difference of the pitch effect is linearly proportional to the difference of sound pressure level between the order of booming sound and its side orders. The relative difference of the pitch effect increases as the frequency increase and arrives to its maximum value around 1200Hz  as shown in Fig. 1. A similar model to this result exists in Ref. 7. However, in this model, the maximum pitch effect is at 700Hz. Moreover, the sound pressure level is not linearly proportional. So we modified this model to fit our results. With considering the pitch effect, the booming index is developed and given mathematically by
Bi = 1/100[25 + 1.25 x [DELTA]p] x [[1 + 0.7[(f/1200 - 1200/f).sup.2].sup.-0.5] (1)
where [DELTA]p is the difference of sound pressure level and f is frequency. The algorithm to realize the booming index is developed based on Eq. (1) and the flow chart is shown in Fig.2. According to this process, we can obtain the booming index of the interior sound if we input the sound file to this algorithm.
The rumbling sound is caused by the bending vibration and torsional vibration of the crankshaft which is excited by the combustion force when the fuel of the engine is fired. The combustion force has a little variation among cylinders. The vibration of the crankshaft is transferred to the interior sound as a structure-borne sound. Therefore, the rumbling sound is caused by the resonances of the structures on the way of transfer path from the engine to car body. There are many resonance frequencies between 200 Hz and 500Hz due to the resonance of the transfer path. The frequency bandwidth of the resonances is wide and this is not a tonal sound. The rumbling sound includes a number of frequencies and is the amplitude modulated sound because the amplitudes of each frequency within bandwidth are difference each other. The rumbling sound can be evaluated objectively by using the modulation of interior sound . It is known that the sound quality of the rumbling sound is correlated to roughness of the interior noise . Roughness is correlated to the weighted form of the modulation of a sound . The rumbling sound can be finally evaluated by using weighted modulation of the sound. In this paper, the modulation of interior noise is calculated based on the harmonic orders of the rotating frequency of the crankshaft as follows :
Step 1. Order tracking analysis. Calculate the amplitude, the phase, the frequency of harmonic order components for the measured interior sound at each rotating speed (rpm) of the engine. The order data is extracted from 0.5 order to 10 order by 0.5 order interval. Because the maximum frequency of the rumbling sound is about 500Hz, it is unnecessary to use an order of more than 10 orders.
Step 2. Filtering. The calculated order data are filtered by the bandpass filter. The frequency range of the band pass filter is from 200Hz to 450Hz.
Step 3. Calculate the envelope of the filtered data. The envelope of the filtered signal is calculated by using Eq. (2).
[mathematical expression not reproducible] (2)
where [A.sub.i] is the amplitude, [f.sub.i] is the frequency, and [[phi].sub.i] is the phase of i- th order.
Step 4. Spectra of envelope. Calculate the spectrum of the envelope of filtered data. In this spectrum, the value at 0Hz is called the DC component. The DC value will be used at step 6.
Step 5. Calculate the spectrum of the envelope weighted by the weighting function. According to theory of roughness , the more the modulation frequency is close to 70 Hz and the carrier frequency approaches at 1 kHz, the more the sound is rough. The weighting function of modulation frequency and carrier frequency is illustrated as shown in Fig. 3.
Step 6. Calculation of the weighted modulation. Applying the weighting function of the step 5 and transformed it into the time domain from the spectrum domain by taking inverse Fourier transform. The root mean square value for this transformed data is called AC. The modulation index can be calculated by dividing the AC value by the DC value calculated in the step 4.
The systematic process for the calculation of the modulation index for the interior sound is presented as shown in Fig. 4.
ORDER DECAY RATIO
Generally, in interior sound of the vehicle, the pleasant sensation tends to be decreased when the powerful sensation increases. Although the two sensations are not exactly in inverse proportion, it is evident that there is a trade-off between pleasantness and powerfulness . Therefore, the relationship among booming index modulation index and pleasant sensation is investigated with the subjective evaluation. According to this investigation, it is found that the booming index is not low correlation with pleasantness and the modulation index is inversely correlated with pleasantness . However, the modulation index is not enough to represent pleasantness alone for the objective evaluation. The spectrum decay method is employed for the objective evaluation of the pleasant sensation. The ratio of spectrum decay means the decay ratio of the sound pressure per octave frequency . In automotive, the integer harmonic orders are the octave frequencies based on the spectrum decay, a new sound metric is developed based on the order components of the engine and it is called "order decay ratio (ODR)". The ODR is calculated through the following procedure. At the first, the sound pressure levels of target order and next two harmonic orders are extracted from the order analysis. For example, in case of in-line four cylinder (I4) engine, if the target order is the 2nd order, other two harmonic orders are the 4th order and the 6th order. In case of the vee six cylinder (V6) engine, if the target order is the 3rd order, other two harmonic orders are 6th order and 9th order. Secondly the decay ratio of the sound pressure level per order is calculated using three extracted order components. Fig. 5 shows the procedure calculating the decay ratio of the sound pressure level per order of interior sound. Finally, the ODR is calculated as the product of the absolute value of the decay ratio and the target order number. This process can be expressed mathematically and is given by
y = [P.sub.1]x + [P.sub.2] ODR = |[p.sub.1]| X the number of target order (3)
where [p.sub.1] is the slope calculated from three extracted orders and [p.sub.2] is a intercept of y.
To evaluate whether the booming index can be used as an objective index of a booming sound, the jury test for interior sounds was performed. The twenty sounds with the different sound pressure level and different frequency were synthesized for the jury test. The frequency per each four sounds among twenty sounds is 50Hz, 100Hz, 150Hz, 200Hz and 250Hz respectively and the difference of sound pressure level for four sounds 6dB, 8dB, 12dB and 24dB respectively. Fig. 6 shows how twenty sounds are synthesized. One sound consists of a harmonic order component and its side order components. The amplitude of all harmonic order components is the same. After the frequency of harmonic order component is selected, the four sounds are synthesized with the difference of sound pressure level between harmonic order and side order component. Fig. 7 shows the correlation between the booming index and the subjective rating for 20 synthetic sounds.
In order to test the utility of the modulation index as a sound metric for the rumbling sound, the twenty sounds are synthesized. The twenty sounds with the different sound pressure level and different frequency are synthesized for the jury test. The carrier frequency of main order per each four sounds among twenty sounds is 200Hz, 250Hz, 300Hz, 350Hz and 400Hz respectively and the difference of sound pressure level between main order and side two orders for four sounds 0dB, 6dB, 12dB and 18dB respectively. The twenty sounds for modulation test are synthesized in the same way as the Fig 6. Fig. 8 shows the correlation between the booming index and the subjective rating for 20 synthetic sounds. In the Fig 8, fc effect means the difference of sound pressure level between harmonic order and side order component is the same but the carrier frequencies are different respectively. Conversely, the dB effect means the carrier frequency is the same but the differences of sound pressure level are different. Finally, 'Random' means sounds with the carrier frequency and the difference of sound pressure level randomly selected.
The developed indexes are applied to four commercial vehicles for the objective evaluation of powerful sensation and pleasant sensation. Four worldwide luxury sedans are selected for this test. They used all V6 engine for the power generation. The vehicles were operated at WOT for around six seconds.
The displacement of these engine is almost similar. Fig. 9 shows the time history type of the interior sounds measured inside cabin of four vehicles. Fig. 10 shows the short Fourier transform for those four time history type corresponding to the signals of the Fig. 9 respectively. In order to design a new interior sound with powerful sensation and pleasant sensation, the objective evaluation index is necessary for the designed sound. Powerful index and pleasant index are developed in this section. For the development of these two objective indexes, the booming index, the modulation index, and order decay ratio for the four measured interior sounds are calculated at each engine speed. The results are presented as shown in Fig. 11 and Fig. 12. In Fig. 11, the vehicle A has booming sensation at low speed around 3500 rpm. The vehicle B shows high level of booming index at 5500 rpm and vehicle D shows high level of booming index at 4800 rpm. However, vehicle C presents low booming sensation at 4300 rpm. Therefore all test vehicles have the booming sensation at difference speed with different booming sensations. This proper level of the booming index is necessary for the powerful sensation of vehicle.
In Fig. 12, the vehicle A has high rumbling sensation at high speed around 5000 rpm. The vehicle B, C, and D show low level of modulation index at 4200 rpm. However, the level of modulation index for all vehicles increases as the engine speed (rpm) increases. Therefore, luxury sedan has the trend that the modulation index increases according to the increment of the engine speed.
Jury test for powerful sensation and pleasant sensation is conducted with four measured interior sounds. These sounds were subjectively evaluated by 23 people for the powerful and pleasant index. Table. 1 illustrates the scale of the powerful sensation and pleasant sensation for the jury test. The results are used for the production of the powerful index and pleasant index. Using the subjective ratings and the sound metrics such as booming index, modulation index and order decay ratio for four measured interior sounds, the multiple linear regress method is applied to obtain two indexes. The formula obtained by multiple linear regress can be used for the objective evaluation of powerful sensation and pleasant sensation of the new designed interior sound and is given mathematically by
Powerful index = 5.18Bi + 84.08Mi - 0.45 (4)
Pleasant index = 0.55ODR - 306.97Mi + 8.5 (5)
where Bi means the booming index, Mi means the modulation index, and ODR means the order decay ratio.
According to these results, the vehicle A is the most powerful while the vehicle C is the best pleasant. However the vehicle A has powerful sensation together with pleasant at low speed under 3000 rpm. Fig. 15 shows the 2 dimensional index for the presentation of pleasantness and powerfulness of the interior sound. The level of x-axis and y-axis designate the mean of the pleasant index and powerful index for all rpm.
In this paper, two indexes which evaluate powerful sensation and pleasant sensations of the interior sound of the vehicle are proposed. These indexes are developed in terms of three sound matrixed such as booming index, modulation index, and order decay ratio. Booming index, which is a sound metric for the evaluation of booming sensation, is proportional to the difference of sound pressure level between the target order and side harmonic orders. This sound metric is also developed by considering the pitch effect related to frequency. The modulation index, which is a sound metric for the evaluation of rumbling sensation, is developed based on the degree of modulation and frequency weighting function. The order decay ratio is the ratio of the sound pressure level per order spectrum. The developed two indexes successfully applied to evaluate the pleasant sensation and powerful sensation of interior sound measured inside cabin of four commercial vehicles. The proposed method can be applied to evaluate the sound quality of actively designed interior sound using ASD method in future.
[1.] Bisping R., "Car Interior Sound Quality: Experimental Analysis by Synthesis," Acta Acustica, 1997.
[2.] Jari K., "Development of a robust and computationally-efficient active sound profiling algorithm in a passenger car," VTT SCIENCE 5, 2012
[3.] Lee S.K., "Objective evaluation of interior sound quality in passenger cars during acceleration," Journal of Sound and Vibration, 2008.
[4.] Lee, S., Chae, H., Park, D., and Jung, S., "Booming Index Development for Sound Quality Evaluation of a Passenger Car," SAE Technical Paper 2003-01-1497, 2003, doi:10.4271/2003-01-1497.
[5.] Kang H.S., Shin T., Park D.C, and Lee S.K., "Quality index of dual shell horns of passenger cars based on a spectrum decay slope," International Journal of Automotive Technology, 2015.
[6.] Borch DZ and Sundberg J., "Spectral distribution of solo voice and accompaniment in pop music," TMH: Quarterly Progress and Status Report, 2002
[7.] Terhardt E., Stoll G. and Seewann M., "Algorithm for extraction of pitch and pitch salience from complex tonal signals," The Journal of the Acoustical Society of America, 1982.
[8.] Zwicker E. and Fastl H., "Psychoacoustics, facts and models," New York: Springer, 1999.
[9.] Lee, S., Kim, B., Chae, H., Park, D. et al., "Sound Quality Analysis of a Passenger Car Based on Rumbling Index," SAE Technical Paper 2005-01-2481, 2005, doi:10.4271/2005-01-2481.
[10.] Janssens, K., Ahrens, S., Bertrand, A., Lanslots, J. et al., "An On-Line, Order-Based Roughness Algorithm," SAE Technical Paper 2007-01-2397, 2007, doi:10.4271/2007-01-2397.
[11.] Kim, S., Chang, K., Park, D., Lee, S., et al. "A Systematic Approach to Engine Sound Design for Enhancing Sound Character by Active Sound Design," SAE Int. J. Passeng. Cars - Mech. Syst. 10(3):in press, 2017, doi:10.4271/2017-01-1756.
Mechanical Engineering, Inha University 100 Inharo, Incheon, South Korea firstname.lastname@example.org
Mechanical Engineering, Inha University 100 Inharo, Incheon, South Korea email@example.com
Hyundai Motor Co. firstname.lastname@example.org
Dong Chul Park
Hyundai Motor Co. email@example.com
This work was supported by Mid-career Researcher Program through NRF of Korea grant funded by the MEST (No. 2015R1A2A1A15052549) and (No. 2016R1A2B2006669)."
Seung Min Lee
Dong Chul Park and Seonghyeon Kim
Hyundai Motor Co.
Sang Kwon Lee
Table. 1. the scale of the powerful sensation and pleasant sensation Score Powerfulness Pleasantness 2-4 Quiet Very unpleasant 4-6 Mild Unpleasant 6-8 Powerful Pleasant 8-9 Noisy Quiet 9-10 Very noisy Too Quiet
|Printer friendly Cite/link Email Feedback|
|Author:||Lee, Seung Min; Park, Dong Chul; Kim, Seonghyeon; Lee, Sang Kwon|
|Publication:||SAE International Journal of Passenger Cars - Electronic and Electrical Systems|
|Article Type:||Technical report|
|Date:||Aug 1, 2017|
|Previous Article:||The Impact of Microphone Location and Beamforming on In-Vehicle Speech Recognition.|
|Next Article:||Analysis of Evaporative and Exhaust-Related On-Board Diagnostic (OBD) Readiness Monitors and DTCs Using I/M and Roadside Data.|