Sound reflection patterns in the classroom of slanted floor

Speech intelligibility (SI) is the measurement of understanding the speech sound by the individuals. In the classroom, modern engineering acoustics has been discussing about the effect of SI such as the academic performance and health of students. Sound reflection in the classroom is one of the factors that affect SI and it is characterized into two types which are early reflections and late reflections. The purpose of this research was to obtain the sound reflection pattern in two classrooms with slanted floor of about the same size. It was done by capturing the sound signal at 12 measurement points in the classroom using Real Wave Analyzer (RWA) and a Smartphone. The sound signals were converted into Power Spectral Density (PSD) using Matlab to analyse the sound reflection patterns. This research also aims to observe the difference of PSD results from which the sound signals were acquired from RWA and Smartphone. It has been found that, the nearer the distance of the ground to ceiling results a higher PSD in both room sample at 1000Hz. Besides, using Smartphone results a higher PSD compared to using RWA at 500Hz. In conclusion, the nearer distance of ground to ceiling in a slanted floor classroom results a higher PSD at 1000Hz


Introduction
Past few years, great attention has been paid discussing about the importance of classroom acoustics.This is due to a greater awareness of the bad effect of poor classroom acoustics to the health of teachers and on student learning [1].According to [2], poor acoustical condition affects the attention and academic performance of the students.
The objective of listening to speech is SI.SI is the overall outcome influenced by various factors.These encompass the actions of both speaker and listener, the characteristics of the room such as its shape and surfaces, as well as the system used for transmitting the speech sound [3].In [4] stated that reflection sound in room is one of the significant factors that influenced SI.According to [2], in classroom, although the majority of students will be able to hear the teacher, not all will be able to comprehend what is being said.This is due to the poor acoustical conditions in the classroom at which the sound listened by students sitting in different locations receive different experience.
When speech was given in a room, sound will travel to the surrounding and reflected to the surfaces such as ceiling, floor and walls [5].This scenario describes the propagation of sound in room at which the originated sound divided into two types of sound, direct and reflective sound [6].Direct sound is the sound transmission travelling in a straight line from the sound source to the listener unaffected by the room's walls or ceiling [5].Besides, reflective sound, also known as reverberation sound gives the meaning of repetitive reflected sound after the sound source stopped by the boundaries [7].The analogy of reflective sound is the listener perceives not only the direct sound from the speech but also the same speech with delayed sound arriving a few seconds later [8].
According to [9], reverberation degrades the SI as it reduces the amplitude modulation that includes the fundamental frequency, blurs spectral signal and flattens formant transitions.However, not all sound reflections give bad impact to SI.It is proven that early sound reflection at 50 milliseconds after the direct sound produced gives the same results as the direct sound increased [4].It was also found out in [10] that the early reflection poses at 0-90 milliseconds delay positively affect the SI as it homogenized with direct sound.Later than 90 milliseconds are consider as late reflections.Based on [11], late reflection is the main factor of detrimental of SI and can be categorized as a kind of noise.
Reverberation time (RT60) has become the objective measurement to analyze the reflections of sound as in [8] and [12].RT60 is the time taken in seconds for sound pressure level to decay at 60dB from the original level of the sound [6].However, this paper observes PSD to analyze sound reflection pattern.PSD describe how power signal is distributed across different frequencies.

Equipment used
Impulsive sound source BAS006 is used as the sound source as shown in Figure 3.This model was chosen because it can create an impulsive noise that can avoid secondary bounce and operates in low frequency energy.These characteristics were suitable to measure reverberation time.It can operate at peak 100-120dB SPL.To capture the sound wave from the sound source created, RWA and a smartphone are used and will be saved in a form of .wavfile.The sound wave will be converted into PSD (in dB/Hz) graph using MATLAB software which will be used for further analysis.Figure 4 below shows the RWA.

Research setup
The sound source created 2 meters in front of the recorded device (RWA and Smartphone) using flapper with 2 feet flapper opening.This work repeated at 12 MP in both RS. Figure 5 and 6 previews the 12 MP in RS1 and RS2 respectively.Table 1 and 2 shows the distance of the MP to the front, back, nearest side wall and from ground to ceiling.These distances need to be measured to analyze the sound reflection patterns in room samples.Laser meter was used to measure these distances.Initially, the flapper sound signals, stored as .wavfiles, were converted from the time-domain into the frequency-domain using Fast Fourier Transform (FFT).This conversion was necessary as the analysis required analysis at specific frequencies.Subsequently, the FFT output was further processed into PSD using the Welch method.These procedures were simulated using Matlab to analyze and define the signals in terms of their amplitudes concerning their respective frequencies, thereby extracting the underlying power spectrum of the signals.

Result and analysis
Two objective measurements are discussed here which are to analyze sound reflection patterns in both RS and to observe the difference of PSD results using RWA and Smartphone in both RS.PSD results in different MP at certain frequency will be analyzed to observe the sound reflection pattern.

Sound reflection pattern
In Figure 8, at RS1, when the PSD result is compared between MP1 and MP2 at 1000Hz, MP2 has a lower PSD at -75dB compared to MP1 at -69dB.It is because the distance of MP2 to front and nearest side wall are 2 meters further as shown in Table 1.However, when comparing MP2 and MP3, MP3 has a higher PSD at -72dB compared to MP2 at -75dB.It is because the distance of MP3 to front and nearest side wall are 2 and 2.3 meters further respectively but the distance from ground to ceiling is 0.2 meter nearer.The same pattern we can see in RS2.In Figure 9, when the PSD result is compared between MP1 and MP2 at 1000Hz, MP2 has a lower PSD at -76dB compared to MP1 at -73dB.It is because the distance of MP2 to front and nearest side wall are 1.6 and 2.2 meters further respectively as shown in table 3.6.However, when comparing MP2 and MP3, MP3 has a higher PSD at -74dB compared to MP2 at -76dB.It is because the distance of MP3 to front and nearest side wall are 1.6 and 1.5 meters further respectively but the distance from ground to ceiling is 0.2 meter nearer.

Difference between using RWA and Smartphone
Next, to observe the difference of PSD results using RWA and Smartphone, PSD were compared in all frequencies.The PSD results in 500Hz shows a consistent pattern which using Smartphone displays a higher value than using RWA at all MP in both RS as shown in Table 3.It can be said that Smartphone has a higher microphone sensitivity towards 500Hz compared to RWA.Table 3. PSD results at 500Hz in RS1 and RS2

Conclusion
The results show that the distance between ground and ceiling has a significant impact on PSD result at 1000Hz.In both room sample, comparing between two points, when the distance between ground and ceiling is constant, the point that is nearer the wall has a higher PSD.However, when the point has a nearer distance between ground and ceiling, the PSD is higher even it is further than the wall.Besides, the PSD results in 500Hz using Smartphone shows a higher value than using RWA at all MP in both RS.Further analysis can be done in this research at different frequencies for deeper analysis.

2. 1 .
Room samplesThis paper uses two room samples which are from lecture rooms in University Teknologi Malaysia (UTM), Skudai.Room sample 1 (RS1) is at block P16 Demo Room 1 in School of Electrical Engineering and room sample 2 (RS2) is at Seminar Room C in School of Civil Engineering.Both room are different by its type of ceiling which RS1 has a flat ceiling while RS2 has a slanted ceiling.Figure1 and 2shows the dimensions of RS1 and RS2 respectively.

Table 1 .
Locations of MP in RS1

Table 2 .
Locations of MP in RS2 2.4.Signal processing Figure 7 outlines the step-by-step process of transforming raw signals of flapper sounds into PSD.