Acoustics Report 3 PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Acoustic Analysis of a New York City Apartment Studio

Alexander Attar
Music and Audio Research Laboratory, Department of Music and Performing Arts Professions New York University - 35 W. 4th Street, Suite 777 New York, NY 100112 Email: aa2429@nyu.edu
AbstractThis paper analyzes the acoustic properties of an apartment bedroom studio, and seeks to define problematic issues with this type of acoustic environment. A number of tests were uses to generate room response using sine sweeps as well as formula calculated data based upon the dimensions of the room. The data was then visualized graphically for comparison. Most prominent issues were observed in the lower end of the frequency spectrum, and additional acoustic non-linearities were present in the frequency range above 2 kHz. Discrepancies were observed between calculated data based upon dimensions vs. the sine sweep tests, and reasons for these discrepancies are speculated.

chance for standing waves considering that the room is in the shape of a rectangular prism with parallel walls. Moreover, this may also cause interference in the higher end of the frequency spectrum in response to the way that the sound waves diffuse. To calculate the data a preliminary goal of this analysis is to excite the system with a signal s(n), measure the response r(n), and use s(n) and r(n) to determine h(n), which is equivalent to the transfer function H(z) = the spectrum of the room deconvolved from the original input signal s(n) [1]. s(n) -> [h(n)] -> r(n) r(n) (*-1) s(n) = [H(z)] (1) (2)

I.

INTRODUCTION

Room acoustics is a practical and diverse area of study. There are many reasons why detailed knowledge of a listening space can be beneficial, but in the context of a studio mixing room it is of particular for understanding where nonlinearities are occurring within the frequency spectrum. This creates the possibility for acoustic treatments that help to linearize the room, and thus allow for a more accurate music production environment. II. MOTIVATION/CONTEXT

In particular, it is useful to pick a signal s(n) that contains a large amount of energy so that measurement noise will not significantly corrupt the measurement results [1]. IV. PROCEDURE

Not everyone who produces music has the consistent luxury of a high-end studio environment that is acoustically designed to match the purpose of the room. With the exponential increase in computational power over the past couple decades, personal computers can now achieve much that used to require big studio rooms filled with expensive gear. This has lead to more people than ever before creating, editing, monitoring and producing their music within the confines of their own homes. Cities such as New York, have become hotbeds for bedroom composers and producers due to the fact that it is an economic solution, but due to the small size of most rooms in the city some considerations must be taken if one wants to improve the acoustic condition of the environment of their home studio. This is the motivation for the analysis of a typical, small and untreated apartment bedroom studio. It is a goal to bring to light specific properties that might be adherent to the type of listening space. Hopefully, the results of the analysis will be useful in the context of one who wishes to measure their own bedroom studio for possible treatment solutions. III. PREDICTED AND CALCULATED DATA

To being the analysis a sine wave sweep had to be created in order to take the impulse response of the room. This was achieved using components of the Transfer Function Toolkit designed in Matlab by Edgar J. Berdahl and Julius O. Smith III [2]. The sine sweep generator utilized four input variables for the user to define including the starting frequency (f1), the ending frequency (f2), the sampling rate (fs), and the length in of the sweep (N) in samples which is converted to time by (2^N)/fs in order to speed up computation. The sine wave generated swept from 20 Hz to 20 kHz at a sample rate of 44.1 kHz over 1 second. The audio file of the sine sweep was then exported to disk. Next, a Pure Data (Pd) patch was designed for reading in the sine sweep in and playback in order to excite the system.

Given the small size of the room it is predicted that most of the acoustic issues will result in the lower end of the frequency spectrum. This is due to the fact that lower frequencies result in longer wavelengths. There is a good

Figure 1. Sine sweep Pure Data patch

The sweep was played in the room and the response was recorded into the Pd patch. Once the patch captured the sine sweep response, the audio was exported to disk. This was then imported into another function from the Transfer Function Toolkit which uses the swept sine excitation stored in the original sine sweep audio file to calculate an inverse filter using the FFT. The rooms response was then output and the results were plotted. Following the observed data from the Matlab and Pd programs, a similar sine sweep procedure was carried out using the Fuzz Measure Pro toolkit, which is a signal analysis application designed for Apple computers. This was done for further comparisons and observations using the cumulative spectral decay and reverberation time plugins that the toolkit offered. The figures generated by both the Matlab/Pd implementations of the sine sweep vs. the Fuzz Measure Pro provide a visual comparison of the room response results:
Figure 4. Matlab Frequency Response

Figure 5. Fuzz Measure Frequency Response Figure 2. Matlab Impulse Response

V.

MEASURED DATA

To begin calculations the room dimensions are needed: Width = 8 ft = 2.4384 meters Length = 10ft = 3.048 meters Height = 8ft = 2.4384 meters Next, it is necessary to determine the average length of time that a sound wave will travel between surfaces in a room. This can be achieved by establishing the mean free path in meters from the following equation [6]: MFP = 4V/S MFP = 4 * 18.123/38.6476 = 1.8757m (3) (4)

Figure 3. Fuzz Measure Impulse Response

There are an infinite number of possible ratios to be used for measuring the reverberation time of sound decay, but the most commonly used ratio is that which corresponds to a decrease in sound energy of 60dB. The gives us what is known as the T

reverberation time. Though T was an appropriate scale for acousticians designing concert halls, it is less useful for analyzing smaller spaces. Shorter decay scales are more when discussing room decay such as a small room studio. So for the purposes of the analysis, we will be using the early decay time (EDT), which will be defined as the time it takes the sound to decrease in sound energy 10dBs in sound energy [6]. Regardless, the T formulas are used first, and the data is converted afterwards. Below are some constants that will be required for carrying out the measurements: Ceiling Surface Area = 2.4384*3.048 = 7.4322432m Floor Surface Area = 2.4384*3.048 = 7.4322432m Walls Surface Area = 64ft + 64ft + 80ft + 80ft = 288ft = 26.7560 m

The time between reflections can be calculated by dividing the mean free path by the speed of sound (344ms) [6]. Time (s) = 4 * 18.123m^3/38.6476m * 344ms = 0.0054527 (5) Sound in the room decays at a rate dependent on the time between reflections and the absorption coefficients of the materials that are within the build of the room. For this analysis, the coefficients for the materials that compose the room were listed in the Howard and Angus text, Acoustics and Psychoacoustics. The coefficients for each of the materials are listed at the following frequencies:

Total Surface Area = 2(wh + lw + lh) = 38.6476m Volume = lwh = 18.123 m3

TABLE I. COEFFICIENTS TO MATERIAL/ FREQUENCY

Rooms built with different areas of material will require calculations to combine the absorption coefficients in a way that accurately reflects their relative contribution. In the Sabine equation: T = 0.161V/S ()) (5)

Given the information from the room measured, the formula can be applied across the frequency spectrum: T = 0.161V/(7.4322432m * @ the variable frequency()) + (7.4322432m * ()) (26.7560m * ()) (8) T Reverb Time (s) = 125 Hz = 1.2043 250 Hz = 1.5959 500 Hz = 1.6922 1 kHz = 2.0447 2 kHz = 2.2825 4 kHz = 2.2825

This is done by multiplying the absorption coefficient of the material by its total area and then adding up the contributions from all the surfaces in the room [6]: (7.4322432m * ) + (7.4322432m * ) (26.7560m * ) (6) Further, the Norris-Eyring Formula allows for consideration of frequency variation of the absorption coefficients [6]: T = -0.161V/S ln(1- ()) (6)

Since the decay time (EDT) is preferred for the purposes of this comparison, the above listed reverb times were divided by 6 in order to analyze the sound decay within the 10dB: 125 Hz = 0.200716 250 Hz = 0.265983 500 Hz = 0.28203 1 kHz = 0.340783 2 kHz = 0.380416 4 kHz = 0.380416

For real rooms we must allow for the presence of a variety of different materials, as well as for their variation of absorption as a function of frequency. This further complicates matters, considering that the different absorption coefficients will have to be combined in a way that accurately reflects their relative contribution in function across the frequency spectrum [6]: T = -0.161V/ Si i()) (7)

Figure 6. Calculated EDT

Figure 8. Matlab Phase Response

Figure 7. Fuzz Measure EDT

VI.

ANALYSIS/DICUSSION
Figure 9. Fuzz Measure Harmonic Distortion

The room response generated by the Matlab/Pd implementation of the sine sweep experiment ended up being remarkably similar to the Fuzz Measure Pro toolkit response. The observation between such agreeable results using the two platforms suggests that the frequency response measurements accurately represent the properties of the room. As predicted, charts generated by both platforms suggest that there are some acoustic issues to account for in the lower end of the frequency spectrum. Particularly, there is a rising bump in the magnitude between 30 Hz up to the peak at 105 Hz. Analysis of the phase also shows signs of harmonic distortion in this area of the spectrum. This suggests that there are most likely some standing waves due to the longer wavelengths of the lower frequencies. A possible solution might require acoustic bass traps to remove the excessive build up of low frequencies.

Figure 10. Cumulative Decay Spectrum

reality there are furnishing within the room environment. It's possible that objects such as beds, curtains, and even the human body could be absorbing the mid-range frequencies. Another, possible fault in the calculated data could involve the actual coefficients of the room materials being different than those suggested by the Howard and Angus text. Finally, it is possible that these discrepancies arise from a difference in the way that EDT was calculated. The process used by the calculations involved dividing the T linearly by 6 to get the EDT. It is possible that the Fuzz Measure Pro toolkit utilizes a different formula that accounts for the logarithmic nature of audio decay. The complexity of these issues would require more work, and so this is an area for future research.

Figure 11. Cumulative Decay Spectrum 2

VII. CONCLUSION To conclude, this analysis brings to light some important issues to consider when working within a small apartment studio. These are starting points for one who wishes to address the problematic nature of this type of acoustic environment. Further research will be aimed more directly at the specifics of dealing with these issues, and particularly focused on affordable solutions for those working within city apartments. REFERENCES
[1] Berdahl, E., & Smith, J. (2008, August 19). Swept Sine Impulse Response Measurement. Retrieved from the Connexions Web site: http://cnx.org/content/m15945/1.4/ E. J. Berdahl and J. O. Smith, Transfer function measurement toolbox, June 2007, http://ccrma.stanford.edu/realsimple/imp meas/. Coefficient Chart. (n.d.). Creative Media Courses for Learning Audio, Video, Film, Games, Animation - SAE Institute. Retrieved July 25, 2011, from http://www.sae.edu/reference_material Farina, Angelo. (2000, Feb.). Simultaneous Measurement of Impulse Response and Distortion with a Swept-Sine Technique. Audio Engineering Society Convention, Preprint 5093 Granados, E. G.(2011, March). Computation of the Impulse Response of small rooms with the time domain Boundary Element Method. Retrieved from.http://upcommons.upc.edu/pfc/bitstream/2099.1/12295/1/MTErnest_Granados.pdf. Howard, D. M., & Angus, J. (2006). Hearing music in different environments. Acoustics and psychoacoustics (3rd ed., pp. 261-347). Oxford: Focal Press

After 105 Hz, the spectrum tapers down to lower fluctuating levels of magnitude, until there is an rise which begins around 2 kHz. This likely suggests that there is interference occurring at the higher frequencies due to the reflections of sound off the untreated parallel walls. The directionality of the high-frequency sound waves in correspondence to the shape of the room and location of the speakers would need to be considered for treatment to solve these issues. A solution would likely involve sound diffusers and absorbers to be installed in strategic locations to minimize the problematic effects. Specifics regarding the proper placement location for these treatments in this type of environment will require further research and experimentation within the space. When comparing the decay time chart of the speculated room response vs. the computer generated decay time chart, some interesting disagreements were observed. The calculated data based on the decay time formulas produced a rising early decay time in correspondence to the rise in frequency, whereas the computer generated data displayed peaks of magnitude at the lower frequencies, a drop in the mid-range, and then a rise beginning again around 2 kHz. There are a variety of reasons that these discrepancies might have occurred. The calculated data was based upon the room being an empty space, but in

[2] [3]

[4]

[5]

[6]

You might also like