A A A Quantification of coin falling sound quality of coin processing machine Motoki Terada 1 Harayama Rikiya 1 Junji Yoshida 1 Osaka Institute of Technology 5-16-1 Omiya, Asahi-ku, Osaka-city, Osaka, 535-8585, Japan ABSTRACT Quietness is essential for living comfortably but various noise exist at around our living place. Coins falling sound of the coin processing machine is one the noise around us. This noise disturbs our conversations, and it causes tired feeing occasionally. The coin falling sound is unsteady sound and have various frequency characteristic during very short time. Hence, not only the sound pressure level but also the quality is considered to be important for the improvement. In this study, we focused on the sound quality of the coin processing machine and investigated the impression through subjec- tive evaluation test. In order to quantify the impressions, various coin falling sounds were prepared and the participants evaluated the “Loudness”, “Pitch”, “Annoyance” and “Tiredness”. In the sound quality analysis, multiple regression analysis was applied to understand which sound factor affects largely to the annoyance and tiredness. As a result, both “Annoyance” and “Tiredness” were observed to be affected by “Loudness” and “Pitch”. Especially, tired feeling was found to be affected more by the pitch of the sound than that of annoyance. From these results, the noise at high frequency band was clarified to be necessary to improve the sound quality. 1. BACKROUND AND PURPOSE Coin processing machines are devices for counting, depositing, and withdrawing coins, and are mainly used in commercial facilities and banks handling large amounts of coins. The operational efficiency is increased by using these machines. On the other hand, radiated noise of the machines in operating may make the workers discomfort and fatigue. Hence, the noise reduction is important. Accordingly, radiated noise analysis and countermeasure were attempted in some previous studies [1, 2]. In a previous study, the main source of the noise was reported to be the coin falling noise into the withdrawal tray [3]. However, which sound characteristic of the coin falling sound affects the discomfort and fatigue feeling was not clarified at present. In this study, we then focused on the sound quality of the coin falling sound and tried to quantify the sound quality to propose the improvement guideline. 2. MEASUREMENT OF COIN FALLING SOUND AND PRESENTED SOUND PREPA- RATION 2.1. Recording of Coin Falling Sound In order to analyze the coin falling sound into the withdrawal tray, we operated a coin processing machine and recorded the radiated noise during the withdrawal process. For the recording, an artificial head microphone (HEAD acoustics HMS2) was placed at 2 m from the machine at a height of 1.55 m. 1 junji.yoshida@oit.ac.jp inter. ales 20: In addition, normal microphone (PCB Piezotronics 130E20) was placed at 1 m from the machine at a height of 1 m. In the recording, only a single coin (1 Euro coin) was ejected to evaluate the withdraw tray characteristic without coin contacting sound. In addition, a simple withdrawal tray model having similar original withdrawal tray characteristic was made for preparing various sound samples for subjective evaluation as shown in Figure 1. Artificial head microphone 2 m Normal microphone (a) Coin processing machine 1 m 1.55 m Coin Coin launch pad 1 m (b) Withdrawal tray model Figure 1: Coin falling sound recording condition using artificial head microphone and normal microphone. Figure 2 shows the specific loudness of the recorded coin falling sound into original withdrawal tray and the tray model according to ISO 532 [4-7]. As shown in the figure, the coin falling sound characteristic into the withdrawal tray model (orange line) was similar with the sound into the original withdrawal tray (blue line). Specific loudness [sone/Bark] 1 1.2 1.4 Original Model 0 0.2 0.4 0.6 0.8 0 2 4 6 8 10 12 14 16 18 20 22 24 Critical band rate [Bark] Figure 2: Comparison of specific loudness of coin falling sound into original withdrawal tray and the tray model. 2.2. Sound Preparation for Subjective Evaluation Test For preparation of the subjective evaluation test, structure of the tray model was modified to prepare various coin falling sounds. At first, four kinds of structure ((a)-(d)) were prepared by using a 3D printer, which had two types of thicknesses and densities as shown in Figure 3. In addition, two types (0.5 and 2 mm) rubber mat were attached inside of the tray. Withdrawal tray model Rubber mat point Figure 3: Various withdrawal tray model structure. In order to evaluate the physical sound pressure characteristic difference among the trays, the sound pressure level (SPL) of 1 Euro coin falling sound was recorded as same as the previous meas- urement. Figure 4 shows the A-weighted SPL comparison in each structure. Figure 4 (a), (b), (c) and (d) are the SPL of coin falling sound to the tray model with thin (0.5 mm), thick (2.0 mm) rubber mat and without mat (Original) in each structure, respectively. 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Original Thin mat Thick mat Original Thin mat Thick mat A-Weighted SPL [dB] A-Weighted SPL [dB] 0 800 1600 2400 3200 4000 4800 5600 6400 0 800 1600 2400 3200 4000 4800 5600 6400 Frequency [Hz] Frequency [Hz] (a)Thickness: 3 mm, Density: 20 % (b) Thickness: 3 mm, Density: 99 % 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Original Thin mat Thick mat Original Thin mat Thick mat A-Weighted SPL [dB] A-Weighted SPL [dB] 0 800 1600 2400 3200 4000 4800 5600 6400 0 800 1600 2400 3200 4000 4800 5600 6400 Frequency [Hz] Frequency [Hz] (c) Thickness: 6 mm, Density: 20 % (d) Thickness: 6 mm, Density: 99 % Figure 4: A-Weighted SPL of coin falling sound into various withdrawal tray. For the increase of the presented sound having various frequency characteristic more in the sub- jective evaluation test, FIR filtering was applied to the recorded coin falling sound on each original tray model without rubber mat. Eight sounds were added by the processing and total 20 sounds were prepared for the subjective evaluation test. Figure 5 (a), (b), (c) and (d) shows the A-weighted SPL of the coin falling sound into the original tray and filtered sounds in each structure. 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 A-Weighted SPL [dB] A-Weighted SPL [dB] A-Weighted SPL [dB] A-Weighted SPL [dB] Original A B Original C D 0 800 1600 2400 3200 4000 4800 5600 6400 0 800 1600 2400 3200 4000 4800 5600 6400 Frequency [Hz] Frequency [Hz] (a)Thickness: 3 mm, Density: 20 % (b) Thickness: 3 mm, Density: 99 % 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Original E F Original G H 0 800 1600 2400 3200 4000 4800 5600 6400 0 800 1600 2400 3200 4000 4800 5600 6400 Frequency [Hz] Frequency [Hz] (c) Thickness: 6 mm, Density: 20 % (d) Thickness: 6 mm, Density: 99 % Figure 5: A-Weighted SPL of falling coins before and after FIR processing. Table 1 (1), (2) and Figure 6 shows the prepared coin falling sounds for the subjective evaluation sound. As shown in Figure 6, many coin falling sound samples having various frequency character- istic could be prepared by the tray model and filter processing. Table 1 (1): Prepared coin fallings sounds into withdrawal tray for subjective evaluation test. No. Loudness [sone] Remarks No. Loudness [sone] Remarks 1 7.91 (a) Original 6 6.02 (b) Original 2 5.64 (a) A 7 5.55 (b) C 3 3.92 (a) B 8 3.53 (b) D 4 8.59 (a) Rubber 0.5 mm 9 7.70 (b) Rubber 0.5 mm 5 9.60 (a) Rubber 2 mm 10 9.54 (b) Rubber 2 mm Table 1 (2): Prepared coin fallings sounds into withdrawal tray for subjective evaluation test. No. Loudness [sone] Remarks No. Loudness [sone] Remarks 11 8.49 (c) Original 16 6.54 (d) Original 12 5.48 (c) E 17 5.95 (d) G 13 5.59 (c) F 18 5.31 (d) H 14 7.97 (c) Rubber 0.5 mm 19 7.00 (d) Rubber 0.5 mm 15 9.30 (c) Rubber 2 mm 20 7.02 (d) Rubber 2 mm 1 1 Specific loudness [sone/Bark] Specific loudness [sone/Bark] Specific loudness [sone/Bark] Specific loudness [sone/Bark] No.6 No.7 No.8 No.9 No.10 No.1 No.2 No.3 No.4 No.5 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24 Critical band rate [Bark] Critical band rate [Bark] (a)Thickness: 3 mm, Density: 20 % (b) Thickness: 3 mm, Density: 99 % 1 1 No.11 No.12 No.13 No.14 No.15 No.16 No.17 No.18 No.19 No.20 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24 Critical band rate [Bark] Critical band rate [Bark] (c) Thickness: 6 mm, Density: 20 % (d) Thickness: 6 mm, Density: 99 % Figure 6: Specific loudness of sound samples for subjective evaluation test. 3. SUBJECTIVE EVALUATION TEST OF COIN FALLING SOUND 3.1. Subjective Evaluation Test Environment Figure 7 shows the experimental equipment and condition for the subjective evaluation test. This subjective evaluation experiment was conducted in an assembled soundproof room (SI- RENTDESIGN) with background A-weighted SPL of less than 30 dB. The evaluation sound was reproduced from a personal computer via a D/D converter (FX-AUDIO's FX-D03J) and through a binaural reproduction device (HEAD acoustics PEQ-V) and headphones (SENNHEISER HD650). Figure 7: Experimental equipment and condition for subjective evaluation test. 3.2. Experimental Procedure In this subjective evaluation test, “Tiredness” and “Annoyance” were used as the overall sound qual- ity evaluation terms, and “Hardness”, “Echo”, “Loudness” and “Pitch” were used as the elemental terms to explain the characteristics of the overall sound quality. In the evaluation test, participants listened to the reproduced coin falling sound and selected one of five categories according to the impression in each evaluation term as shown in Figure 8. In addition, for the further analysis, integer number was attached in each category e.g., “Very tired”: 5, “Slightly tired”: 4, “Neither”: 3, “Slightly comfortable”: 2 and “Very comfortable”: 1 in case of tiredness evaluation. (5 points judgement) Very Slightly Neither Slightly Very (5 points judgement) Very Slightly Neither Slightly Very Comfort Tired Muffle Echo Quiet Annoying Loud Small Hard Soft High pitch Low pitch 1 2 3 4 5 1 2 3 4 5 Figure 8: Evaluation sheet and the category for the subjective evaluation test of coin falling sound. This test was carried out in each session including 20 sounds evaluations, and each participant carried out five sessions in total. The sound presentation order was randomized to reduce the presen- tation order effect. Total 13 Japanese students in their 20-30 years, who gave informed consent ap- proved by the university's ethics committee, participated in the experiment. Therefore, total number of trials of the experiment became 1300 (13 participants × 20 evaluation tones × 5 sessions = 1300 times), and the total number of evaluations per sound was 13 participants × 5 times = 65 times. 3.3. Subjective Evaluation Score in Each Term Figure 9 shows the average score of the presented 20 sound samples in each term of “Annoyance”, “Tiredness”, “Pitch” and “Loudness”. Annoying Annoyance Tiredness Tired 5 5 4 4 3 3 2 2 Comfort Quiet 1 1 No,1 No,2 No,3 No,4 No,5 No,6 No,7 No,8 No,9 No,1 No,2 No,3 No,4 No,5 No,6 No,7 No,8 No,9 No,10 No,11 No,12 No,13 No,14 No,15 No,16 No,17 No,18 No,19 No,20 No,10 No,11 No,12 No,13 No,14 No,15 No,16 No,17 No,18 No,19 No,20 High pitch Pitch Loudness Loud 5 5 4 4 3 3 2 2 Low pitch Small 1 1 No,1 No,2 No,3 No,4 No,5 No,6 No,7 No,8 No,9 No,10 No,11 No,12 No,13 No,14 No,15 No,16 No,17 No,18 No,19 No,20 No,1 No,2 No,3 No,4 No,5 No,6 No,7 No,8 No,9 No,10 No,11 No,12 No,13 No,14 No,15 No,16 No,17 No,18 No,19 No,20 Figure 9: Averaged score of the presented sound in each evaluation term. As the first analysis, ANOVA (analysis of variance) was conducted for each evaluation term and the result showed that the significance probability of each term was under 0.05 and evaluation terms were judged to be suitable for this evaluation. In addition, the coin falling sound of structural modi- fication with rubber mats were observed to be evaluated as soft and low-pitched sounds (No. 3, 7, 8, 12, 13, 17, and 18) in the loudness and pitch evaluations. Furthermore, these sounds were also tended to be evaluated as low “Tiredness” and “Annoyance”. 4. CHARACTERISTICS OF OVERALL SOUND QUALITY OF COIN FALLING SOUND 4.1. Relationship Between Overall Sound Quality Metrics and The Elemental Metrics Here, multiple regression analysis was conducted for understanding the characteristics of overall sound quality metrics (“Tiredness” and “Annoyance”). As the objective variables, “Tiredness” and “Annoyance” score were used and “Loudness” and “Pitch” score were used as the explanatory vari- ables in this analysis. Equations 1, 2 and Figure 10 (a), (b) show the regression equations and the relationship between subjective and calculated overall sound quality scores. Noting that other ele- mentary terms (“Echo” and “Hardness”) were not used in this analysis because the relationship be- tween these terms and the overall metrics were not significant. Tiredness = −0.283 + 0.592 𝐿𝑜𝑢𝑑𝑛𝑒𝑠𝑠+ 0.547 𝑃𝑖𝑡𝑐ℎ (1) Annoyance = −0.988 + 1.116 𝐿𝑜𝑢𝑑𝑛𝑒𝑠𝑠+ 0.221 𝑃𝑖𝑡𝑐ℎ (2) Vertical and horizontal axes in Figure 10 show the subjective and calculated scores by using Equa- tion 1 and 2, respectively. In addition, R in the figure represents the correlation coefficient. As shown in both figures, the coefficients were very high at 0.979 in “Tiredness” (Figure 10 (a)) evaluation and 0.991 in “Annoyance” (Figure 10 (b)) evaluation. 5 5 Measured value Measured value 4 4 3 3 2 2 R = 0.979 R = 0.991 1 1 1 2 3 4 5 1 2 3 4 5 Predicted value Predicted value (a) Tiredness (b) Annoyance Figure 10: Relationship between subjective and calculated scores of overall sound quality evaluation terms. From these analytical results, both “Tiredness” and “Annoyance” are found to be expressed well by using “Loudness” and “Pitch.” And in case the sound falling sound is soft and low pitch, the sound seems to be low tired and low annoyance. Here, standardized partial regression coefficients in each regression equation were calculated as shown in Table 2 to evaluate the difference between overall sound quality metrics of “Tiredness” and “Annoyance” in detail. Table 2: Standardized partial regression coefficients for “Tiredness” and “Annoyance”. Loudness Pitch Tiredness 0.466 0.615 Annoyance 0.827 0.234 As indicated in this table, the coefficient of pitch in “Tiredness” evaluation was much larger than that of “Annoyance” evaluation. On the other side, the coefficient of “Loudness” in “Tiredness” was smaller than the coefficient in “Annoyance” evaluation. This indicates that the tired feeling of the coin falling sound was especially affected largely by the high pitch sound. 4.2. Quantification of Overall Sound Quality of Coin Falling Sound Here, we attempted to express from the recorded sound pressure signals to estimate the tiredness and annoyance. We then employed psychoacoustic metrics of Loudness and Sharpness, which have been reported to show good adaptation to auditory loudness and sharpness (pitch) [8], to show the subjec- tive “Loudness” and “Pitch”. Single regression analysis was carried out and obtained the equation to show the subjective “Loudness” and “Pitch”. Equations 3 and 4 show the equations and Figure 11 shows the relationship between subjective and calculated “Loudness” and “Pitch”. In the equations, 𝑋 L and 𝑋 S show the calculated overall loudness and sharpness, respectively. Loudness = 2.054 + 0.184𝑋 L (3) Pitch = 0.384 + 2.33𝑋 S (4) 5 5 Measured value Measured value 4 4 3 3 2 2 R = 0.813 R = 0.651 1 1 1 2 3 4 5 1 2 3 4 5 Predicted value Predicted value (a) Loudness (b) Pitch Figure 11: Relationship between subjective and calculated loudness and pitch of the coin falling sound. The correlation coefficient of both elemental terms was high at 0.651 in (a) and 0.813 in (b). Sub- sequently, the overall sound quality metrics of “Tiredness” and “Annoyance” were expressed quan- titatively using Equations 1-4. Equations 5 and 6 shows the equation for “Tiredness” and “Annoy- ance” calculated from psychoacoustic metrics (Loudness and Sharpness). Tiredness = 0.410 + 0.133𝑋 L + 1.688𝑋 S (5) Annoyance = 0.517 + 0.176𝑋 L + 1.310𝑋 S (6) By using the upper equations, “Tiredness” and “Annoyance,” can be estimated from the recorded coin falling sound signal. Figure 12 shows the relationship between subjective overall sound quality metrics and estimated value from the recorded sound pressure signals using Equations 5 and 6. 5 5 Measured value Measured value 4 4 3 3 2 2 R = 0. 822 R = 0.909 1 1 1 2 3 4 5 1 2 3 4 5 Predicted value Predicted value (a) Tiredness (b) Annoyance Figure 12: Relationship between estimated and measured overall sound quality metrics. As shown in upper figures, the correlation coefficients were very high 0.909 in the “Tiredness” evaluation and 0.822 in the “Annoyance” evaluation. In addition, the sharpness (high pitch) of the sound was found to affect largely especially on the tired impression and the reduction of the high pitch noise is considered to be effective way to relieve the tired feeling by the coin falling sound. 5. VALIDATION OF THE COIN FALLING SOUND QUALITY QUANTIFICATION Here, we conducted subjective evaluation test again using different coin falling sounds to verify whether the obtained equations can express the overall sound quality metrics (“Tiredness” and “An- noyance”) well or not. For the validation sound quality evaluation test, three coin falling sound sam- ples were prepared by using FIR filtering. In the filtering, preparing low or high sharpness sound. Table 3 shows the calculated loudness and sharpness of the three sound samples. Table 3: Prepared coin falling sound for the verification test. Evaluation sound Loudness [sone] Sharpness [acum] Original 10.0 1.49 Low frequency reduction (High pitch) 9.23 1.72 High frequency reduction (Low Pitch) 8.31 1.25 Before carrying out subjective evaluation test, the overall sound quality metrics of “Tiredness” and “Annoyance” were calculated using Equations 5 and 6. Black points in Figure 12 show the esti- mated value in each sound quality matric. As shown in the figure, low frequency reduction sound (High pitch sound) was indicated to be most tiredness and annoyance sound. Subsequently, subjective evaluation test of “Tiredness” and “Annoyance” was conducted through a parried comparison method. Three evaluation pairs were prepared, and the evaluation was repeated three times in each metric. Five males aged 20-30 years participated, hence, total evaluation number in each pair became 5 persons × 3 times = 15 times. As the subjective score of each presented sound sample, the number of selections in paired comparison was counted in each overall sound quality metrics. Blue and orange bars in Figure 13 shows the selection number of each sound samples in each overall sound quality metrics of “Tiredness” and “Annoyance.” 30 30 4.7 4.7 25 25 4.5 Number of selection Number of selection 4.5 Estimated score Estimated score 20 20 4.3 4.3 15 15 4.1 4.1 10 10 3.9 3.9 5 5 3.7 3.7 0 3.5 0 3.5 Original High Low Original High Low (a) Tiredness (b) Annoyance Figure 13: Comparison of the number of selections in paired comparison and the calculated value from the quantification equation in each overall sound quality metric. As shown in the figure, the estimated tiredness and annoyance score of each presented sound (black points in Figure 12) are observed to have very high correlation with the number of selection (blue and orange bars). These results indicate that the equation for expressing overall coin falling sound quality of “Tiredness” and “Annoyance” in Equations 5 and 6, actually could estimate the feeling very well. And the overall sound quality were contributed by loudness and pitch of the sound. However, the influence of pitch of the sound (sharpness) on the “Tiredness” was larger than that of “Annoyance”. Accordingly, high-frequency noise reduction is effective way to reduce tiredness of workers near the coin processing machine. 6. SUMMARY In this study, the sound quality of coin falling sound from the withdrawal tray was investigated to propose the improvement guideline. At first, 20 coin falling sounds were prepared for subjective evaluation test. Subsequently, the characteristics of overall coin falling sound quality “Tiredness” and “Annoyance” were found to be expressed well by using psychoacoustic metrics of loudness and sharpness through multiple regression analysis. This result also revealed that “Tiredness” and “An- noyance” were differently affected by pitch of the sound (sharpness), and “Tiredness” was especially affected largely by sharpness. And the tendency was again confirmed through the verification sub- jective evaluation test. The above results indicate that high-frequency noise reduction is effective way to reduce tiredness of workers using coin processing machine. 7. REFERENCES 1. Hosaka, R. Low-noise technology and sound quality improvement for home appliances. The Acoustic Society of Japan , 62(10) , 744-749, (2006). In Japanese 2. Yamamoto, K. Methods for reducing interior noise of rail vehicles. Inst. of Noise Control Engi- neering of Japan , 31(5) , 368-373, (2007) . In Japanese 3. Yamamoto, T., Matsukawa, T. & Yoshida, J. Noise analysis and countermeasures for coin pro- cessing machine using calculated loudness. Proceeding of Mechanical Engineering Congress , 2020 Japan , p. 320. Nagoya, Japan, September2020. In Japanese 4. ISO 532 Acoustics-Method for calculating loudness level, 1975. 5. Suzuki, Y., Takeshima, H., Measurement and International Standardization of Human Equal Loudness Curves. Journal of the Institute of Electrical Engineers of Japan , 124(11) , 715-718 (2004). In Japanese 6. Zwicker E. & Fastl, H. Psychoacoustics: Facts and Models, 3rd Edition, Springer, 2006. 7. Zwicker E. Subdivision of the audible frequency range into critical bands (Frequenzgruppen). The Journal of the Acoustical Society of America , 33(2) , 248 (1961). 8. Takada, M. How to calculate sound quality evaluation index and examples of application, The Acoustical Society of Japan, 75(10) , 582-589 (2019). In Japanese Previous Paper 373 of 769 Next