A Study Of Speech Recognition Of The Elderly In Speech-shaped Noise And Reverberation

With the aging of population in China,the demand for residential buildings for the elderly is increasing.As the hearing gradually decreases with age,it is more difficult for the elderly than the young to understand the voice in the noise and reverberation environment,which leads to the barriers of speech communication among the elderly,resulting into loneliness,inferiority,loss of self-confidence and other psychological problems.A good architectural acoustic environment is particularly important for the speech communication of the elderly.But in terms of the existing interior acoustic environment design of residential buildings for the elderly,architects lack of relevant codes,standard guidance and reference.Therefore,this work studies the Chinese speech intelligibility of interior space in residential buildings for the elderly,and discusses the influence of reverberation time(RT),background noise level,signal-to-noise ratio(SNR)and other acoustic environmental factors on the speech recognition of the elderly,which can provide reference for the acoustic environment design of residential buildings for the elderly,and is of great significance to improve the speech communication of the elderly in their residential buildings.In this work,the objective acoustic parameters and binaural room impulse responses of the listening positions in rooms with different reverberation time were obtained by room acoustic simulation;then the binaural room impulse responses were convoluted and mixed with the test signals and speech-shaped noises recorded in the anechoic room;finally,the subjective Chinese speech intelligibility was performed by the elderly of different ages and hearing levels according to a certain SNR through headphones.The relationship between subjective evaluation results and objective acoustic parameters of the elderly with different ages and hearing levels under different RT and different background noise levels was studied.The results are as follows.The subjective Chinese speech intelligibility of the elderly aged 60-69 with normal hearing was evaluated under different background noise levels(40d BA and 55 d BA),different RT(0-2.1s)and different SNR(-5?12d B).Under the same RT and SNR conditions,the average score of the elderly with normal hearing is 6.3% lower than that of the young when the background noise level is 40 d BA,and 9.9% lower when the background noise level is 55 d BA.In order to achieve the basic speech communication ability,under the 40 d BA background noise level and RT of 0.6s conditions,the needed SNR of the elderly aged 60-69 with normal hearing is10.4d BA,and that of the elderly aged over 70 with mild hearing loss is 21.6d BA;Under the background noise level of 55 d BA and RT of 0.6s conditions,the needed SNR of the elderly aged 60-69 normal hearing and aged over 70 with mild hearing loss are 6.0d BA and 14.2d BA,respectively.The elderly are more sensitive to the changes in environmental conditions(reverberation and reverberation plus noise)than the young adults.The lower the background noise level is,the smaller the score difference between the elderly and the young.Reducing the background noise is helpful to improve the speech communication of the elderly.This study explores the relationship between the scores of Chinese speech intelligibility and the speech transmission index(STI)and speech intelligibility index(SII)of the elderly aged 60-69 and over 70.The results show that there is a good correlation between the scores of Chinese intelligibility and STI,SII.In order to achieve the same speech intelligibility as the young,the STI of the elderly aged 60-69 and the elderly aged over 70 should be increased by0.058 and 0.419 respectively compared with the young in 40 d BA speech-shaped noise,and0.072 and 0.282 respectively in 55 d BA speech-shaped noise.Because the hearing loss is considered in the calculation of SII,the difference of SII between the elderly and the young is less than the difference of STI.The speech reception threshold(SRT)of elderly people with different hearing levels was tested under different RT and different background noise levels.The results show that SRT increases with the increase of age and hearing loss.SRT is highly linearly correlated with RT in different age groups and hearing level groups.Under the background noise level of 40 d BA and 55 d BA,the slopes of the linear fitting relationship between SRT and RT increase with age and the average pure tone hearing.In the RT range of 0 s to 2.15 s,the increment of SNR needed by the elderly increase with the increase of RT,hearing loss and age.Under the background noise of 40 d BA,the SNR of the elderly with normal hearing,mild hearing loss and moderate hearing loss should be increased by 6.62 d BA,10.08 d BA and 17.15 d BA,respectively,in order to reach 50% of the speech intelligibility required by SRT,while the SNR of the elderly aged60-69 and over 70 should be increased by 6.98 d BA and 14.24 d BA,respectively.Under 55 d BA background noise,the values are 3.85 d BA,6.21 d BA and 11.89 d BA,3.92 d BA and 9.76 d BA,respectively.Under the same RT condition,the SRT value of the elderly under 40 d BA background noise was higher than that of 55 d BA background noise level,but its speech sound pressure level was lower than that of 55 d BA background noise level.Under the condition of low background noise level,it is more beneficial to improve the speech intelligibility of the elderly by increasing the speech sound pressure level and improving the SNR.On the basis of SRT test results,STI and SII corresponding to SRT were calculated,and the relationships between SRT and STI,SRT and SII under different reverberation conditions were discussed.The results show that the more severe the hearing loss or the older the elderly,the greater the STI is needed to achieve the sentence intelligibility(50%)required by SRT.Under the condition of 40 d BA background noise level,the STI and SII calculated from the SRT values of the elderly under different RT conditions decrease linearly with the increase of SRT.Under 55 d BA background noise level,the SRT values of the elderly with normal hearing,mild hearing loss and aged 60-69 under different RT conditions can be approximately expressed by STI of 0.447,0.500 and 0.450,respectively;the SRT values of the elderly with normal hearing,mild hearing loss and moderate hearing loss,as well as 60-69 years old and over 70 years old under different RT conditions can be approximately expressed by SII of 0.428,0.457 and 0.482,and 0.433 and 0.469,respectively;while the STI of the elderly with moderate hearing loss and the elderly aged over 70 under different RT conditions decrease linearly with the increase of SRT.In order to ensure a good speech communication environment,the elderly need a room acoustic environment more than the young,which should be with short reverberation time and low background noise to realize daily normal speech communication.Architects should take some acoustic treatment measures in the design of buildings for the elderly,such as sound absorption,noise reduction and sound insulation,to reduce the background noise level and RT,and improve the indoor SNR,so as to improve the indoor speech intelligibility of residential buildings for the elderly.