Listening difficulty of public announcement at subway platform with long reverberation time

Eunsung Song 1 , Suhong Kim 2 , Jongkwan Ryu 3 Chonnam National University 77, YongBong-ro, Buk-gu, Gwangju, Republic of Korea

ABSTRACT The subway platform is a unique space where the reverberation time (RT) is extended depending on the distance from the sound source. Moreover, various noises are being occurred at the subway platform such as the train approaching noise, crowd noise, and so on. The purpose of this study is to determine which factor affects the listening difficulty of announcement through listening test based on the acoustical data from on-site measurements. The results of listening test showed that the difference in listening difficulty according to the types of noise sources, S/N ratio, and RT was significant. In particular, the difference between RTs of 2.2 s and 3.6 s showed the largest difference of up to 65.1 % in listening difficulty rating. In addition, the influence of noise type on listening difficulty rating was found in the condition of RT 2.2 s, but not in RT 3.6 s. Announcement combined with the crowd noise showed highest listening difficulty rating in the case of better acoustical condition.

1. INTRODUCTION

It was found that underground subway platforms, which account for 67.5 % of the total platform types in the world [1], have mostly reverberation time longer than 2.0 s due to the long-shaped structure and the reverberant finishing materials. In addition, it is a unique space that requires clear announcements. However, it was reported that the signal-to-noise ratio (S/N ratio) of the announcement was distributed lower than 10 dBA in some platforms [2,3]. Besides, due to the high correlation between the reverberation time and the listening difficulty (LD) [4,5,6], it is necessary to evaluate the listening difficulty in a space with a long reverberation time such as a subway platform. Moreover, it is necessary to check the effect of the spectrum of noise on listening difficulty because the actual listening environment for announcement in the platform is exposed to various noise sources as well as background noise. Therefore, the purpose of this study was conducted in order to understand which factors affect the listening difficulty of the announcement in a situation composed of actual noise sources, the reverberation time, and S/N ratio of a subway platform.

1 218272@jnu.ac.kr 2 tfh0409@naver.com 3 jkryu@jnu.ac.kr

2. METHOD

2.1. Listening difficulty test This listening test is to evaluate how difficult it is to hear the announcement sound in a situation such as a passenger waiting for a train in an underground subway station. A total of 32 subjects with normal hearing evaluated listening difficulty of announcement ( ‘This train is bound for XX station’ ) in the environment of 4 noise types, 3 speech-to-noise ratio (S/N ratio) and 2 reverberation times as shown in Table 1. In order to reduce word dependence, only words with a final consonant, a characteristic of Phonetic Balanced word (PB word), were selected, and a total of 192 words (name of station) were selected by conducting a dictionary word familiarity evaluation for subjects. Listening difficulty (LD) was evaluated on a 4-point scale: 'Not difficult, 'A little difficult', 'Fairy difficult', and 'Extremely difficult'. Analysis was calculated as the ratio of the number of responses from 2 (A little difficult) to 4 (Extremely difficult) excluding 1 (Not difficult). The entire experimental procedure consisted of a total of 16 sessions, and 2 sets were composed according to RT. The words presented in each session through Sennheiser HD 600 and were randomly arranged. Table 1: Experimental design of listening difficulty test

Evaluation type Noise type S/N ratio

RT(500Hz)

Announcement

[dBA]

[s]

level [dBA]

Train operation sound (TA1)

0 / 10 / 20 2.2 / 3.6 70 Train brake sound (TA2)

Speech intelligibility

Crowd sound(A+C)

Ambient sound(A)

2.2. Sound sources Figure 1 shows the time history of the four types of noise and the announcement. Train arriving noises (TA1 and TA2) showed a non-stationary tendency, but ambient (A) and crowd (C) noise showed a stationary tendency. Figure 2 indicates the frequency characteristics of all sound sources. For ambient noise (A), the sound pressure level ( SPL ) decreased with increasing frequency range over 250 Hz, but the low frequency band rather increased in the 63 Hz band. In the case of ambient and crowd noise (A+C), it shows a similar trend to ambient noise (A), but decreases in the 63 Hz band. For train arriving noise (TA1), SPL in mid-frequency band (500 Hz ~ 1 kHz) was the highest, and SPL in high- frequency band was the lowest among sound sources. In the case of train arriving noise (TA2), which is characterized by a train break sound, SPL in low-mid frequency band was the lowest among sound sources, but high SPL was found at 2k Hz to 4k Hz. In the case of the announcement, which is a male voice, it shows the most similar tendency to the ambient and crowd noise (A+C).

(a) Non-stationary noise (b) Stationary noise Figure 1: The time history of each noise sources for two reverberation times (A: ambient, C: crowd,

TA: train arriving)

(a) 2.2 s (b) 3.6 s Figure 2: The frequency characteristics of each noise sources and announcement for two

reverberation times (A: ambient, C: crowd, TA: train arriving) 3. RESULTS

3.1. S/N ratio and noise type As shown in Figure 3 (a), the listening difficulty decreased as the S/N ratio increased. First, for the difference in S/N ratio as an average over noise types, the difference between 0 dBA and 10 dBA was 36.4 %. Whereas the difference between 10dBA and 20dBA was found to be relatively low at 10%. Moreover, the listening difficulty rating at S/N ratios of 0 to 20 dBA differs depending on each noise source as shown in Figure 3 (b). In the case of poor S/N ratio condition (0 dBA), there were a large variation in listening difficulty rating among noise types, and ambient noise (A) showed the highest level. However, small variation was observed in the case of better S/N ratio condition (20 dBA). In particular, ambient and crowd noise (A+C) was found to be difficult to listen at all S/N ratios. Maximum difference in listening difficulty rating among S/N ratios for subway arriving noise (TA1 and TA2), ambient and crowd noise (A+C), and ambient noise source (A) was 39 %, 46 %, and 62 % respectively.

100

SNR 0 dBA

Listening difficulty rating [%]

SNR 10 dBA

SNR 20 dBA

80

60

40

20

0

A+C

TA1

TA2

A

Type of the noise

(a) (b)

Figure 3: Listening difficulty rating for each S/N ratio (a) and noise type (b) on a reverberation time

(2.2 s)

3.2. RT with S/N ratio and noise type Listening difficulty rating for each reverberation time on the various conditions of noise type and S/N ratio was plotted in Figure 4. As shown in Figure 4, the listening difficulty rating in RT of 3.6 s was significantly longer than in RT of 2.2 s for all S/N ratios. The difference in listening difficulty rating between RTs was lager in the case of greater S/N ratio. The difference in the listening difficulty rating as an average over noise type for 0, 10, 20 dBA was 19 %, 46 %, and 53 %. In addition, the difficulty rating difference between RTs varied with the type of noise. When the S/N ratio is 0 dBA, the difference for the subway noise source (TA1, TA2) was 28.3% and for the ambient and crowd noise source (A, A+C) was 9%. In the S/N ratios of 10 dBA and 20 dBA, there were little variation with noise type.

(a) 0 dBA (b) 10 dBA (c) 20 dBA Figure 4: Listening difficulty rating for each reverberation time on the various conditions of noise

type and S/N ratio

4. CONCLUSIONS

This study conducted the listening difficulty test using sound source with the long reverberation time under three S/N ratios by announcements and various noise sources on the subway platform. It was found that the listening difficulty rating was significantly different according to the type of noise source, S/N ratio, and reverberation time. In particular, listening difficulty rating between the reverberation times (2.2 s and 3.6 s) was around 53 % at the S/N ratio of 20 dBA, which showing the largest difference among the influencing factors. In addition, it was found that the announcement with ambient and crowd (A+C) noise source was the most difficult to hear in the better condition, where

RT is 2.2 s and S/N ratios are 10 dBA and 20 dBA. This is judged to be like the effect of information masking, which has a large effect under good sound environment conditions. In the future, we plan to proceed evaluations in a wider range of reverberation time considering elderly people with hearing loss. 5. ACKNOWLEDGEMENTS

This research was supported by Research Program through the National Research Foundation of Korea (NRF) (No.2019R1A2B5 B01070413).

6. REFERENCES

1. World metro figures, 2018, Advancing public transport, statistics brief, pp. 2 ~ 6 2. Song, H. S., Kim, J. H., Lee, S. M., Kim, S. H. and Ryu, J. K., 2020, Field study of signal and

guide sound for the elderly in subway station. In: INTER-NOISE and NOISE-CON Congress and Conference Proceedings. Institute of Noise Control Engineering, p. 1618-1624 3. Song, E. S., Kim, S. H., Kim, J. H., Lee, S. M., & Ryu, J. K., 2022, The Distribution Characteristic

of Noise and Signal Guide Sound in Subway Platform, Transactions of the Korean Society for Noise and Vibration Engineering, 32(1), 11-21 4. Morimoto, M., Sato, H., & Kobayashi, M. ,2004, Listening difficulty as a subjective measure for

evaluation of speech transmission performance in public spaces. The Journal of the Acoustical Society of America, 116(3), 1607-1613. 5. Sato, H., Morimoto, M., Sato, H., & Wada, M. ,2008, Relationship between listening difficulty

and acoustical objective measures in reverberant sound fields. The Journal of the Acoustical Society of America, 123(4), 2087-2093. 6. Kim, Y. H., & Soeta, Y., 2013, Effects of reverberation and spatial diffuseness on the speech

intelligibility of public address sounds in subway platform for young and aged people, ASA