acoustics

72
Universit¨ at des Saarlandes Spoken Language Systems Development of Test and Measurement Software for Electroacoustic Parameters on Hearing Aids Submitted to Saarland University Faculty of Natural Sciences and Technology I Department of Computer Science in partial fulfillment of the requirements for the degree of Master of Science in Computer and Communications Technology by Mohamed Imran Noor Mohamed Supervisor Prof. Dr. Dietrich Klakow Advisor Dipl.-Ing. Oleg Fallmann Dipl.-Ing. Uwe Nauerz Reviewers Prof. Dr. Dietrich Klakow Prof. Dr.-Ing. Chihao Xu January 2015

description

automation

Transcript of acoustics

  • Universitat des SaarlandesSpoken Language Systems

    Development of Test and MeasurementSoftware for Electroacoustic Parameters

    on Hearing Aids

    Submitted to Saarland University

    Faculty of Natural Sciences and Technology I

    Department of Computer Science

    in partial fulfillment

    of the requirements for the degree of

    Master of Science in Computer and Communications Technology

    by

    Mohamed Imran Noor Mohamed

    Supervisor

    Prof. Dr. Dietrich Klakow

    Advisor

    Dipl.-Ing. Oleg Fallmann

    Dipl.-Ing. Uwe Nauerz

    Reviewers

    Prof. Dr. Dietrich Klakow

    Prof. Dr.-Ing. Chihao Xu

    January 2015

  • Eidesstattliche Erklarung

    Ich erklare hiermit an Eides Statt, dass ich die vorliegende Arbeit selbststandig verfasst

    und keine anderen als die angegebenen Quellen und Hilfsmittel verwendet habe. Ich

    erklare hiermit an Eides Statt, dass die vorliegende Arbeit mit der elektronischen Ver-

    sion ubereinstimmt.

    Statement in Lieu of an Oath

    I hereby confirm that I have written this thesis on my own and that I have not used any

    other media or materials than the ones referred to in this thesis. I hereby confirm the

    congruence of the contents of the printed data and the electronic version of the thesis.

    ........................................................ .................................................

    (Ort und Datum / Place and Date) (Unterschrift / Signature)

    Einverstandniserklarung

    Ich bin damit einverstanden, dass meine (bestandene) Arbeit in gedruckter und elek-

    tronischer Form in die Bibliothek der Informatik aufgenommen und damit veroffentlicht

    wird. Ich erklare dass mit einer derartigen Veroffentlichung keine Rechte Dritter verletzt

    werden.

    Declaration of Consent

    I consent to make my thesis (with a passing grade) accessible to the public by having it

    added to the library of the Computer Science Department in electronic and in printed

    form. I declare that a publication of this kind does not infringe any rights of third

    parties.

    ........................................................ .................................................

    (Ort und Datum / Place and Date) (Unterschrift / Signature)

    ii

  • For their unconditional love and support, I owe this thesis to my family and almighty.

    - Imran

    iii

  • Acknowledgements

    I would like to express my sincere gratitude to my supervisor Prof. Dr. Dietrich Klakow

    for a wonderful learning opportunity he provided to me. I feel deeply indebted to the

    way he patiently let me learn from my own mistakes. With his excellent supervision, I

    have learned and improved a lot about the process of doing scientific research. Because

    of his invaluable guidance, I have been able to groom myself as a better contributor

    towards research. I consider myself very fortunate to have worked with him.

    I would like to thank my mentor Oleg Fallmann for his invaluable guidance. I feel

    deeply indebted for his strong advises, encouragements and instructions. He has not

    only assisted me in completing this thesis, but also helped me to broaden my attitude

    towards research, and in developing my personality.

    I am highly grateful to Prof. Dr.-Ing. Chihao Xu for reviewing my thesis. I

    would also like to thank advisor Uwe Nauerz for his technical guidance throughout the

    thesis. I would like to take this opportunity to also thank all the members of Cetecom

    organization for their support.

    I am also very grateful to my friends and classmates for their company and moral

    support. I learned a lot from them in last two years which is a very important factor for

    successful completion of my thesis. Finally, I thank my almighty for all the good deeds.

    iv

  • Abstract

    For an effective test systems, having a reliable and efficient automated test and mea-

    surement solutions is of critical important. Especially for evaluation of hearing aids

    having an accurate and time efficient calibration using an equalization algorithm is the

    paramount objective to satisfy stringent requirements posed by IEC 60118 standards.

    In the context of this Masters thesis we aimed to develop an automated test and

    measurement software to evaluate electro-acoustical parameters on hearing aids by im-

    plementing a novel equalization algorithm to compensate unfavorable room effects on

    the grounds of characterization of room acoustics.

    The novelty of our developed equalization algorithm is the algorithm equalizes bad

    room acoustics with an accuracy of +/ 0.5 dB and time taken for equalization is below1 hour. Suitable experiments on room characterization finds the optimized positions

    to place microphone and loudspeaker inside the room which significantly reduces the

    reflections by adding acoustic absorbers.

    The implemented novel equalization algorithm outperforms the default proprietary

    solutions in terms of accuracy and time efficiency. We had developed a solution which

    produces a reliable and accurate results for all sound pressure levels (90. 80. 70, 60

    dB) except for less than 1% of all data points in 50 dB which is still considered accept-

    able. Hence, the developed solutions elevates the capability of next generation test and

    measurement systems by satisfying performance metrics set by IEC standards.

  • Contents

    Acknowledgements iv

    Abstract vi

    List of Figures x

    List of Tables xii

    Abbreviations xiii

    1 Introduction 1

    1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

    1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

    1.3 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

    2 Problem Formulation 4

    2.1 Need for Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

    2.2 Equalization by R&S K-7 Software . . . . . . . . . . . . . . . . . . . . . . 6

    2.3 Equalization by Cetecom software . . . . . . . . . . . . . . . . . . . . . . 7

    2.4 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

    2.5 Plan of Attack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

    3 Acoustics: Overview 10

    3.1 Fundamentals of Sound . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

    3.2 Propagation of Sound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

    3.3 Sound Level and the Decibel . . . . . . . . . . . . . . . . . . . . . . . . . 12

    3.4 Acoustic Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

    3.5 Room Acoustics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

    3.6 Absorption on Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

    3.7 Early Reflections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

    4 Methodology 16

    4.1 Software Design and Architecture . . . . . . . . . . . . . . . . . . . . . . . 16

    4.1.1 SCPI: Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

    4.1.2 VISA Library . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

    4.1.3 Event Driven Programming . . . . . . . . . . . . . . . . . . . . . . 19

    viii

  • Contents ix

    4.1.4 MVC: Design Pattern . . . . . . . . . . . . . . . . . . . . . . . . . 20

    5 Characterization of Room Acoustics 23

    5.1 Room Acoustic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

    5.2 Reverberation Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

    5.2.1 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

    5.2.2 Practical Measurements . . . . . . . . . . . . . . . . . . . . . . . . 25

    5.2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

    5.3 Room Impulse Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

    5.3.1 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

    5.3.2 Practical Measurements . . . . . . . . . . . . . . . . . . . . . . . . 28

    5.3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

    5.4 Room Frequency Response Analysis . . . . . . . . . . . . . . . . . . . . . 30

    5.5 Room Mode Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

    5.6 Room Noise Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

    5.6.1 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

    5.6.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

    5.7 Evaluation of Room Characteristics . . . . . . . . . . . . . . . . . . . . . . 35

    5.7.1 Reverberations and Impulse Response . . . . . . . . . . . . . . . . 35

    5.7.2 Loudspeaker Evaluation . . . . . . . . . . . . . . . . . . . . . . . . 36

    5.8 Proposal for Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . 38

    5.8.1 Adding Absorbers . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

    5.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

    6 Equalization 42

    6.1 Equalization Design and Working . . . . . . . . . . . . . . . . . . . . . . . 42

    6.1.1 Programming Decision . . . . . . . . . . . . . . . . . . . . . . . . . 42

    6.1.2 Theory and Block Diagram . . . . . . . . . . . . . . . . . . . . . . 43

    6.2 Setup and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

    6.3 Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

    6.3.1 Verification at 50 dB SPL . . . . . . . . . . . . . . . . . . . . . . . 49

    6.3.2 Statistical Accuracy at 50 dB SPL . . . . . . . . . . . . . . . . . . 51

    6.4 Algorithm Runtime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

    6.5 System Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

    6.5.1 Comparison of Field Frequency Response . . . . . . . . . . . . . . 53

    6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

    7 Conclusion and Future Work 55

    Bibliography 57

  • List of Figures

    2.1 Sound source calibration by R & S; K-7 software . . . . . . . . . . . . . . 6

    2.2 Sound source calibration by Cetecom(Minimum Linearization) . . . . . . 7

    2.3 Sound source calibration by Cetecom(Maximum Linearization) . . . . . . 8

    3.1 Condensation and rarefaction effect of sound . . . . . . . . . . . . . . . . 11

    3.2 Sound travel inside a room . . . . . . . . . . . . . . . . . . . . . . . . . . 14

    4.1 Example SCPI command explaining device operation . . . . . . . . . . . . 17

    4.2 VISA hierarchy with different communication bus protocols . . . . . . . . 18

    4.3 Event driven programming with several processes . . . . . . . . . . . . . . 19

    4.4 Model View Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

    4.5 MVC- UML based design of modules . . . . . . . . . . . . . . . . . . . . . 21

    5.1 Flow chart explaining RT30 measurement . . . . . . . . . . . . . . . . . . 26

    5.2 RT30 plot inside acoustic chamber . . . . . . . . . . . . . . . . . . . . . . 27

    5.3 Ideal RIR plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

    5.4 Flow chart explaining RIR measurement . . . . . . . . . . . . . . . . . . . 28

    5.5 RIR inside the acoustic chamber . . . . . . . . . . . . . . . . . . . . . . . 29

    5.6 Room dimension for frequency response measurement . . . . . . . . . . . 30

    5.7 Frequency response measurement inside acoustic chamber . . . . . . . . . 31

    5.8 Room modes inside acoustic chamber . . . . . . . . . . . . . . . . . . . . . 33

    5.9 Noise explanation with system theory . . . . . . . . . . . . . . . . . . . . 33

    5.10 Noise inside the acoustic chamber . . . . . . . . . . . . . . . . . . . . . . . 34

    5.11 RT-30 plot inside the dead room . . . . . . . . . . . . . . . . . . . . . . . 35

    5.12 RIR plot inside the dead room . . . . . . . . . . . . . . . . . . . . . . . . 36

    5.13 Frequency response plot inside dead room . . . . . . . . . . . . . . . . . . 37

    5.14 Standard Deviation plot for KS Digital loudspeakers . . . . . . . . . . . . 38

    5.15 RIR: room without absorbers . . . . . . . . . . . . . . . . . . . . . . . . . 39

    5.16 RIR: Room improvement with absorbers . . . . . . . . . . . . . . . . . . . 39

    5.17 RT-30: Room improvement with absorbers . . . . . . . . . . . . . . . . . 40

    6.1 Equalization algorithm design: adaptation from M.Sc seminar . . . . . . . 44

    6.2 Equalization module flow chart . . . . . . . . . . . . . . . . . . . . . . . . 47

    6.3 Best case setup inside acoustic chamber . . . . . . . . . . . . . . . . . . . 48

    6.4 Equalization results for 90 and 80 dB SPL . . . . . . . . . . . . . . . . . . 49

    6.5 Equalization results for 70 and 60 dB SPL . . . . . . . . . . . . . . . . . . 50

    6.6 Equalization results for 50 dB SPL . . . . . . . . . . . . . . . . . . . . . . 50

    6.7 Verification at 50 dB SPL . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

    6.8 Statistical Deviation with +/ 0.5 dB . . . . . . . . . . . . . . . . . . . . 52x

  • List of Figures xi

    6.9 Algorithm run time as size of data increases . . . . . . . . . . . . . . . . . 53

    6.10 Field of frequency response: using R &S system . . . . . . . . . . . . . . . 53

    6.11 Field of frequency response: using novel equalization algorithm . . . . . . 54

  • List of Tables

    5.1 Room Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

    5.2 Room Mode Calculation for Cabin . . . . . . . . . . . . . . . . . . . . . . 32

    xii

  • Abbreviations

    ISO International Organization for Standardization

    IEC International Electrotechnical Commission

    SPL Sound Pressure Level

    B & K Bruel and Kaer

    R & S Rhode and Schwarz

    SCPI Standard Commands Programmable Instruments

    IVI Interchangeable Virtual Instruments

    VISA Virtual Instruments Software Architecture

    COM Communication

    MVC Model View Controller

    RT Reverberation Time

    DSP Digital Signal Processing

    RIR Room Impulse Response

    xiii

  • Chapter 1

    Introduction

    In the next-generation test and measurement systems, automated test equipment and

    solutions are critical components that determines the effectiveness and efficiency of the

    overall test system. The automated test solutions are envisioned to offer a variety of ad-

    vantages over the current systems in terms of flexibility, cost effectiveness and reliability.

    However, when we consider evaluation of hearing aids performance using automated test

    solutions the electro-acoustical measures form the basis [1]. The realization of an effec-

    tive electro-acoustical test and measurement system for hearing aids requires a rigorous

    interplay between acoustic knowledge and the understanding of the existing standards,

    as effects of electro-acoustic characteristics has an elusive link to behavioral aspects for

    the hearing aid users [2].

    1.1 Background

    As one of the prerequisite to achieve, a fast, reliable and effective electro-acoustical mea-

    surement is calibration of the acoustic components such as microphone and loudspeak-

    ers. Although, the techniques to calibrate microphones is trivial, yet current systems

    and methods for calibrating loudspeakers with a connection to a microphone located at

    a listening area in a room are often limited to manual calibrations. The development

    of a self-calibrating solutions for loudspeaker includes specific features to adjust speaker

    characteristics based on the effects generated by operating the loudspeaker in the room.

    Under this scenario, equalization for room acoustics plays a vital role to produce desired

    1

  • Chapter 1. Introduction 2

    sound field inside the room [3]. The term Equalization has broad meaning, in specific,

    we consider to adjust or linearize the frequency components produced inside the acoustic

    room.

    As an example, the microphone picks up a test signal generated by the loudspeaker

    and the system uses the test signal to determine the loudspeaker frequency response.

    Thus, the frequency response is analyzed and characterized. Based on the analysis, the

    system generates parameters to compensate unfavorable effects inside the room with the

    help of an efficient equalization algorithm. This constitutes the key to automate tasks

    which, in certain cases, requires accuracy and time-efficiency.

    The International Electro-technical Commission(IEC) has published several stan-

    dards for various applications within hearing aids. Of particular importance is IEC

    60118, which constitutes the standard for the measurement procedure for hearing aids

    [4]. Hence, to reproduce precise electro-acoustical characteristics on hearing aids the

    accuracy in the equalization as technique for calibration is to be +/ 0.5 dB(decibel).For deviations higher or lesser than 0.5 dB seems to alter the frequency response on

    the hearing aids. Similarly, having an effective measurement time to evaluate hearing

    aids also needs to be considered for practical feasibility. Although, the description of

    accuracy and time efficiency exactly does not correlate to IEC standards. However,

    to achieve standard compliance the automated test solution need the aforementioned

    requirements for effective evaluation of the device under the test.

    1.2 Motivation

    Various equalization techniques have been proposed for implementing the room response

    compensation such as DSP(Digital Signal Processing) based techniques by Karjalainen

    et al. [5] and other techniques involving modal equalization [6]. Yet, in the context of the

    thesis, the development of an equalization algorithm considers the stringent requirements

    in accuracy of +/ 0.5 dB and measurement time efficiency within 1 hour as a basis tosatisfy IEC standards. Similarly, our requirement is to develop an equalization based

    on a non post-processing algorithm as extra signal processing will influence hearing aids

    characteristics. This requirements brings up novelty and challenges in the thesis to

    automate test and measurement systems especially for hearing aids.

  • Chapter 1. Introduction 3

    The present thesis project is conducted at the Cetecom ICT Services GmbH, Saar-

    brucken with a specific innovation objective to investigate and implement efficient room

    response equalization, making the automation of electro-acoustical measurements reli-

    able using existing tools.

    1.3 Outline

    Chapter 2 explains the problem addressed in the Masters thesis and the researchquestions that are posed.

    Chapter 3 describes the basic theory of acoustics and room acoustics.

    Chapter 4 explains the methodology to design software.

    Chapter 5 explains the characterization of room acoustics.

    Chapter 6 explains the design and implementation of the equalization algorithmwith system testing and verification.

    Chapter 7 explains the the summary of our work with possible opportunities ofthe future work are discussed.

  • Chapter 2

    Problem Formulation

    Development of an efficient test and measurement solution for hearing aids measurements

    are critical. The entire application development involves synchronization and triggering

    of measurement routines to process measurement data and error management. Taking

    all these parameters into account, the development of an instrument control software

    presents several challenges.

    This chapter describes the problem that is addressed in the present Masters thesis

    and the research questions that are posed.

    2.1 Need for Calibration

    Acoustic measurements often involves measurement of sound pressure level or the gen-

    eration of a sound field, or both. Typical examples are noise measurements, loudspeaker

    measurements, microphone measurements and measurements on systems like hearing

    aids and mobile phones. A sound level meter, for example, is supposed to directly mea-

    sure the sound pressure level in dB SPL (dB referred to 20 Pa sound pressure). If an

    audio analyzers(such as Rhode & Schwarz UPV) is used for this purpose, the sensitivity

    of the measuring microphone Sm

    Sm = Vm/p (2.1)

    4

  • Chapter 2. Problem Formulation 5

    has to be determined, wherein Vm is the RMS value of the microphone output voltage

    and p is the RMS value of the acoustic pressure on the membrane. From the equation 2.1,

    the sound pressure is obtained by dividing the measured voltage by this sensitivity value.

    The measurement by which the value of the microphone sensitivity is obtained is known

    as microphone calibration. The sound pressure level for the microphone calibration is

    generated with an acoustic calibrator (such as B & K) which generates a sine wave signal

    with defined frequency (commonly 1 kHz) and defined sound pressure (usually 1 Pa or

    10 Pa). As measurement microphones are small and have a well-defined mechanical

    structure, the sensitivity is frequency-independent within a 4k Hz till 20k Hz frequency

    range. Therefore, calibration at a single frequency is sufficient.

    However, certain measurements requires sound generation using loudspeakers in a

    defined sound pressure at a certain point. Thus, loudspeakers sensitivity is a measure

    of Sound Pressure Level (SPL) at a specified distance for a specified input signal in RMS

    voltage, typically at one or more specified frequencies (often 300, 400, 500, 600 Hz or

    the average of these).

    As a loudspeaker has a far more complex mechanical structure than a measuring

    microphone, and radiation effects additionally influence the loudspeaker sensitivity de-

    pending on the frequency and on the location of the measurement point, the sensitivity

    of a loudspeaker usually is frequency dependent. For this reason the calibration of a

    loudspeaker consists of two steps:

    Measurement of the absolute sensitivity at one frequency

    Measurement of the frequency response relative to this frequency.

    In order to generate a defined sound pressure level at the measurement point, the

    generator output sound pressure has to be set to the desired sound pressure divided by

    the loudspeaker sensitivity, and to be corrected by an efficient inverse frequency response

    as known as equalization.

    This process of producing required sound pressure level from generator or loudspeaker

    involving equalization is known as loudspeaker calibration. The Equalization plays a

    crucial role in finding desired sound pressure level (dB SPL) for an applied voltage(dBV)

    at different frequencies.

  • Chapter 2. Problem Formulation 6

    Since calibration is the first procedure that needs to be conducted before any measure-

    ments. Having accurate and efficient calibration is important especially for conformance

    testing procedure as explained by hearing aids standards such as IEC 60118.

    2.2 Equalization by R&S K-7 Software

    To manage and record calibration values for acoustic devices such as sound source, mi-

    crophone and ear simulator automatically by means of suitable routines R & S developed

    K-7 software. However, while performing calibration of the sound source the K-7 soft-

    ware misses the accuracy to equalize the sound pressure well within the range of +/- 0.5

    dB SPL, which is a stringent requirement from the hearing aid standards IEC 60118.

    Figure 2.1: Sound source calibration by R & S; K-7 software

    As an illustration, we did equalization for the sound source (loudspeaker) on sound

    pressure levels 50, 60, 70, 80, 90 dB SPL using K-7. As shown in Figure.2.1 after 3000

    Hz till 9000 Hz the equalization of the sound pressure levels (in dB SPL) is not accurate

    and it deviates up to 1 dB SPL in all sound pressure levels (50 dB SPL till 90 dB SPL).

    Therefore, the results shows that equalized values does not satisfy requirements set

    by standards. With this accuracy, further measurements involving hearing aids would

    produce less accuracy and the whole compliance cannot be certified for the device under

    test as the test solution is not accurate enough adhering the standard requirements.

  • Chapter 2. Problem Formulation 7

    Hence, the entire equalization solution becomes a bottle neck for hearing aids testing

    within the acoustic lab. Therefore, we need a novel research solution to solve the equal-

    ization problem within the standards maintaining both accuracy and time efficiency.

    2.3 Equalization by Cetecom software

    To solve the equalization problem, Cetecom developed an algorithm to linearize the

    sound pressure values. Initially, they developed a software with accuracy greater than

    +/- 0.5 dB SPL and named as Minimum Linearization algorithm which equalizes the

    sound pressure level in a well defined time (less than 30 minutes). Yet, the algorithm is

    less accurate as shown in Figure.2.2

    Figure 2.2: Sound source calibration by Cetecom(Minimum Linearization)

    However, to increase the accuracy of the algorithm they developed Maximum Lin-

    earization algorithm by increasing the iterative procedure used in the software for much

    longer cycles. Hence, the new developed solution is accurate but not time efficient

    (greater than 1 hour) as shown in the Figure.2.3

    Being time efficient is not a requirement from the standard for performance evalua-

    tion. Yet, for typical hearing aid measurements we need a time budget to start and stop

  • Chapter 2. Problem Formulation 8

    test procedure well within a time frame. As without time efficient solution the software

    becomes impractical.

    Under this scenario, we need an equalization algorithm which linearizes the sound

    pressure levels within an certain accuracy (+/- 0.5 dB) and be time efficient (typically

    less than 1 hour)

    Figure 2.3: Sound source calibration by Cetecom(Maximum Linearization)

    2.4 Research Questions

    In the context of the thesis, several research questions are posed concerning the develop-

    ment of accurate and time efficient equalization algorithm to correct bad room acoustics

    for further hearing aid measurements.

    What are the basic characteristics and quality assessment can be done for theacoustic chamber to develop an efficient equalization algorithm?

    What possible improvements can be achieved feasibly based on the detailed anal-ysis for the acoustic chamber?

    Which is the proposed equalization solution for overcoming the challenges in termsof accuracy and time efficiency?

    What are the potential limitation and bottlenecks for the developed solution inthe context of hearing aid measurements that occur while satisfying the stringent

    requirement of IEC hearing aid standards?

  • Chapter 2. Problem Formulation 9

    The theoretical research and implementation of solution was conducted in the context

    of the present Masters thesis. It aims to answer the above mentioned questions in a

    detailed manner. The thesis involves development of efficient equalization solution for

    hearing aid measurement by detailed room acoustic analysis.

    2.5 Plan of Attack

    To solve and realize our problem formulation in a systematic way. We structure our

    approach into four particular stages namely:

    Software Design and Architecture: The set of operations defined by the measure-ment instruments should be compatible with our software development. Hence,

    a brief understanding is required to map our software into the system. Here we

    understand the resources and libraries used for mapping in Chapter 4

    Characterization of our acoustic chamber: To provide scientific solution to ourproblem, we need to understand the properties of our room with respect to acous-

    tics. Thus, behavioral measurements will lead us to develop modules efficiently

    and accurately corresponding to our scenario in Chapter 5

    Equalization for room non-linearities: An equalization algorithm is designed andimplemented in a way that the performance of the algorithm is guaranteed for its

    efficiency and accuracy. Therefore, statistical accuracy and algorithm run time

    is taken into consideration with high priority to verify the algorithm for its best

    effort performance in Chapter 6

    System Testing and Verification: The implemented equalization algorithm is testedfor its system functionality and evaluated with the default scenario using R&S K-7

    software in Chapter 6

  • Chapter 3

    Acoustics: Overview

    In the present chapter, the basic theory of acoustics and in particular room acoustics

    terminology are presented. This chapter covers the essential theory acquired during the

    phase of thesis based on which characterization of the room for efficient equalization is

    developed.

    3.1 Fundamentals of Sound

    Sound is defined as vibration transferred through a medium. For example, if an in-

    strument such as loudspeaker sounds, the vibration from the instrument is transmitted

    into another medium which is the air. From a physical point of view, the instruments

    vibrations are pushed as a pattern of condensation and rarefactions into the air. This

    pattern is further transmitted to the surrounding air and hence a longitudinal wave is

    created by a vibrating object. Thus, the wave carries the pattern of vibrations through

    the air at approximately 343 meters per second(m/s) at 20 C.

    To illustrate, how sound waves are produced and why they are longitudinal, consider

    the vibrating diaphragm of a loudspeaker. When the diaphragm moves outward, it

    compresses the air directly in front of it, as shown in Figure.3.1. This compression

    causes the air pressure to rise slightly. The region of increased pressure is called a

    condensation, and it travels away from the speaker at the speed of sound.

    After producing a condensation, the diaphragm reverses its motion and moves inward,

    as in Figure.3.1 (part b). The inward motion produces a region known as a rarefaction,

    10

  • Chapter 3. Acoustics: Overview 11

    Figure 3.1: Condensation and rarefaction effect of sound

    where the air pressure is slightly less than normal. Following immediately behind the

    condensation, the rarefaction also travels away from the speaker at the speed of sound.

    3.2 Propagation of Sound

    From basics of sound, we understood that sound is a vibration that propagates through

    a medium. Hence, sound can be created or transmitted only in a medium such a gas,

    liquid or solid. Moreover, the particles present in the medium must disturb the wave to

    move from one place to another. Thus, sound cannot exist in vacuum.

    To understand sound propagation it is essential to know physical properties of sound

    such as frequency and wavelength. The frequency of a waveform is how many full cycles

    of the waveform we hear in a second. Its value is one divided by the period, where the

    period is the amount of time it takes for the waveform to complete a full cycle as defined

    in cycles per second or Hertz (Hz). In general, wavelength is a measure of the distance

    between repetitions of peaks in a waves cycle. It is defined as :

    = V/f (3.1)

    Where is the wavelength and V is the phase speed of the wave and f is wave

    frequency.

  • Chapter 3. Acoustics: Overview 12

    Like any other waves, sound waves undergo certain behaviors when it encounters the

    end of the medium or obstacles. Possible behaviors include reflections off the obstacle,

    diffraction around the obstacle, and transmission (accompanied by refraction) into the

    obstacle or new medium. For instance, diffraction happens when the waves bends around

    small obstacles.

    3.3 Sound Level and the Decibel

    As compared to other physical units and measurements in acoustics, the intensity of

    sound is measured in decibel (dB). Understanding decibel and its related units will help

    us to explore into the area of acoustics and sound. In principle, the sound intensity is

    defined as the sound power per unit area. The intensity I is measured as Decibel(dB)

    scale.

    I(dB) = 10 log10[I/I0] (3.2)

    Where I0 = 1012Watts/m2,

    By definition, decibel is the ratio of a given intensity to the threshold of hearing

    intensity, where this threshold takes the value 0 dB. The logarithm involved in the

    formula is just the power of ten of the sound intensity as a multiple of the hearing

    intensity. For example, if I/I0 is 104 then equivalent sound intensity is 40 dB. Similar

    to the sound power we can measure sound level as L in dBV by applying voltages as v

    and v0, where v0 is referenced to 1 V.

    By rearranging,

    L(dBV ) = 20 log10(v/v0) (3.3)

    V = V0 10L(dBV )/20 (3.4)

    The above relations between the applied voltage (in V) and produced sound pressure

    (in dBV) is important to derive right sound level.

  • Chapter 3. Acoustics: Overview 13

    3.4 Acoustic Components

    A microphone is an acoustics-to-electric transducer which converts the sound pressure

    into an electric signal. There are different varieties of microphone available but for our

    practical measurements a precise pressure sensing condenser microphones are used. It

    uses a constant electrical charge to convert the diaphragm displacement into an analog

    electrical signal.

    A loudspeaker is a device used to convert an electric signal into acoustic waves. Loud-

    speaker is one of the important variables which may alter the frequency response of the

    chamber which needs to be linearized. In principle, ideal loudspeakers produce acoustic

    waves that are a linear transformation of the electrical input signal. Thus, it could be

    seen as a LTI (Linear time-invariant) system. However, it is also well known that loud-

    speakers produce non-linearity which can be modeled mathematically [7]. Even though

    loudspeaker manufacturers takes the non-linearity measures into consideration. Yet, the

    actual cause and symptoms for non-linearity in different loudspeakers is well discussed

    by Klippel, Wolfgang [8]. Also, compensation for the non-linearity in loudspeaker is a

    topic of its own research as investigated on M.Sc Seminar.

    In thesis, while developing equalization algorithm the acoustic characteristics pro-

    duced from the loudspeaker is considered extensively before calibration is done. Yet,

    internal characteristics of loudspeaker is not considered. Hence, we model loudspeaker

    and other acoustics components such as amplifiers as black box model during develop-

    ment.

    3.5 Room Acoustics

    Consider the propagation of sound in an enclosed room. The sound conducting medium

    is bounded on all sides of wall, floor and ceiling. These room boundaries are not ideal

    surfaces, they absorb a portion of the sound energy impinging on them, and reflect the

    rest. The absorbed energy is either converted into heat, or transmitted to the outside

    of walls. These numerous reflected components is responsible for what is known as the

    acoustics of a room.

  • Chapter 3. Acoustics: Overview 14

    If an acoustic wave hits a hard flat surface, it is reflected back (in Figure.3.2) into

    the room due to the large difference in the acoustic impedance between the air and the

    material of the hard surface.

    Figure 3.2: Sound travel inside a room

    The phenomenon of having reflected sounds in an enclosure is known as Reverberation.

    The time taken for the entire activity is known is Reverberation Time (in Section 5.2)

    3.6 Absorption on Surfaces

    Real surfaces absorb a portion of the energy of the incident sound wave. Absorption

    usually depends on the frequency. The absorption coefficient is the ratio between the in-

    tensity Ii of the incident wave and the intensity difference between incident and reflected

    wave Ir:

    = (Ii Ir)/Ii (3.5)

    An open window in a room has more or less 100% absorption ( = 1 ) at least for

    frequencies where the wavelength is small compared to the dimensions of the window.

    Hence, choosing an appropriate absorbing material will reduce reflections happening

    inside the room.

  • Chapter 3. Acoustics: Overview 15

    3.7 Early Reflections

    A sound impulse starts from a sound source traveling at equal speed in all directions. In

    different distance from the sound source at different directions, the impulse hits a hard

    surface of a wall or other obstacle in the room and is reflected.

    At a certain point of observation, the direct sound traveling the shortest distance from

    the source to the point of observation arrives first. One after the other, reflections arrive

    from different directions, delayed according to the additional distance the impulse had

    to travel due to its detour. If the sound intensity at the point of observation is plotted

    over time, a so called reflectogram(Room Impulse Response) is obtained. Based on the

    Room Impulse Response, the exact positions of reflections happening inside the room

    can be found, which is conducted as an experiment in Section 5.3.

  • Chapter 4

    Methodology

    In the following sections, we explain the software design and architecture used in our

    solution. Although it is not the aim of this thesis to study and understand the technical

    details of a real platform and instruments used. Yet, in order to implement a running

    prototype, having a brief knowledge on architecture and tools used is a requirement.

    4.1 Software Design and Architecture

    In context of the thesis, to meet all the technical and operational requirements with

    feasibility, while optimizing essential quality attributes such as performance we need to

    define a structured solution. Hence, software design and architecture encompasses the

    required set of significant structure to develop good software.

    4.1.1 SCPI: Overview

    For effective communication between the PC and instrument via IEEE 488 bus stan-

    dard commands is required. Thus, the SCPI- Standard Commands for Programmable

    Instruments was defined by IVI (Interchangeable Virtual Instruments) foundation [9]

    to have a generic command to communicate between instruments. In principle, SCPI

    defines the roles of instruments and controllers in a measurement system.

    The following are advantages of using SCPI:

    16

  • Chapter 4. Methodology 17

    Usability - Easy to query the applications as it uses template style which makes itgeneric.

    Consistency - From the remote programming standpoint, it offers a consistent com-munication between instruments of the same class with good functional capability.

    Maintainability - Easy to maintain and extend commands.

    Let us understand how SCPI commands are used in our software with an example to

    appreciate its advantages.

    Example: Sub Routine : IntializeSystemSetttings

    Figure 4.1: Example SCPI command explaining device operation

    In the Figure 4.1, a subroutine uses SCPI commands to control the instruments

    such as in line 2, the commands instructs the instrument to put the display ON while

    measuring data and then it switches the device output to ON state. Finally, it asks for

    the device to return Boolean values as result of operation completion of the previous

    executed commands. Thus, the above commands instruct the instruments with less effort

    for us to memorize specific commands. Furthermore, it brings advantages to formulate

    commands based on the structure provided by the standards.

  • Chapter 4. Methodology 18

    4.1.2 VISA Library

    Virtual Instrument Software Architecture, commonly known as VISA, is a widely used

    input-output(I/O) API (Application Program Interface) to communicate with instru-

    ments and the PC. Generally, for test system development we require a software ar-

    chitecture that allows easy interchangeability of instruments between different systems.

    Thus, using open industry standard software architectures such as VISA, we are able to

    create systems with interchangeable test instrumentation [10].

    The VISA standard library includes specifications for communication between re-

    sources over device specific I/O interfaces such as GPIB (General Programmable In-

    strumental Bus) or IEEE 488 bus. For our application, we used VISA architecture with

    .NET framework where the library interacts over the Microsoft communication (COM)

    technology.

    Figure 4.2: VISA hierarchy with different communication bus protocols

    In the Figure.4.2, the VISA library stands as gateway to communicate between instru-

    ments with different bus protocols. In our scenario, we use GPIB over 488.2 specification

    for communication. The diagram exemplifies VISA library offering as :

    An abstraction to differentiate between buses,

    Incorporates different bus protocols and classes with the good functional capability,

  • Chapter 4. Methodology 19

    Finally, it provides a unified API to communicate between test equipment-regardlessof communication bus.

    Hence, we employed the VISA- software architecture to communicate between our

    instruments efficiently.

    4.1.3 Event Driven Programming

    To test any application there is a well defined sequence (events) or standard procedure by

    which system testing is needs to be done continuously. With this test sequential model,

    we need to develop software that is reusable, maintainable and extensible corresponding

    to various sequence (events) while testing. These features will also enable the software

    to be adaptable to plug out old implementations and plug in new implementations with

    ease. So, there is need to develop a system corresponding to events as it elevates the

    possibility to quickly test the research ideas for real implementation.

    Figure 4.3: Event driven programming with several processes

    Hence, to achieve the above mentioned paradigm we need Event-Driven programming.

    It is a programming technique that corresponds to different events based on the action

    intended. In an event-driven application, there is generally a main loop that listens for

    events, and then triggers functions when one of those events is detected. It can be best

    understood with the help of an example.

    Let us look at the Figure.4.3. a user executes a single process indicated in Green

    circle, the single process calls several other sub-process indicated as Red and Violet

  • Chapter 4. Methodology 20

    circles and the whole event is repeated until the successful final event or when a dead

    lock happens Orange circle.

    In an event-driven architecture, information can be propagated in near-real-time

    throughout as a highly distributed environment. It promotes low run time of an ap-

    plication which is a requirement for our application development in Section 2.4. Also,

    this type of programming techniques helps us to handle tasks sequentially according to

    the test plan.

    4.1.4 MVC: Design Pattern

    Based on the event driven programming paradigm explained in the above paragraphs,

    Model-View-Controller (MVC) is the right choice for the software design pattern. As

    this pattern separates the modeling of the domain, the presentation, and the actions

    based on user input into three separate classes [11]:

    Briefly, the MVC contains:

    Model: The model manages the behavior and data of the application domain,responds to requests for information about its state (usually from the view), and

    responds to instructions to change state (usually from the controller).

    View: The view manages the display of information.

    Controller: The controller interprets the mouse and keyboard inputs from the user,informing the model and/or the view to change as appropriate.

    Figure.4.4 depicts the structural relationship between three objects. It is important

    to note that both the view and the controller depends on the model. Hence, a Model-

    View-Controller is a fundamental design pattern for the separation of user interface logic

    from requirement logic corresponding to events.

    One of the key advantage of using the MVC design pattern is testability. Testing

    components becomes difficult when they are highly interdependent, especially with user

    interface components. These types of components often require a complex setup just to

    test a simple function. Worse, when an error occurs, it is hard to isolate the problem to a

  • Chapter 4. Methodology 21

    Figure 4.4: Model View Control

    Figure 4.5: MVC- UML based design of modules

  • Chapter 4. Methodology 22

    specific component. This is the reason why separation of concerns is such an important

    architectural driver.

    Thus, MVC separates the concern of storing, displaying, and updating data into three

    components that can be tested individually. So, the use of MVC design pattern is quiet

    applicable for our event driven development.

    Based on the understanding of MVC-design pattern, we implemented each modules or

    class using this model as it was feasible. As an outcome, the Figure.4.5 has each module

    designed using MVC-pattern. Finally, all components or modules are aggregated into a

    single unit.

  • Chapter 5

    Characterization of Room

    Acoustics

    This chapter covers the characterization of the room with suitable experiments.

    5.1 Room Acoustic Analysis

    From our problem statement in Section 2.4, room characteristics will give more insight to

    understand rooms behavior with respect to sound. Therefore, understanding essential

    room qualities will help us to design effective equalization solution with scientific goals.

    The following sections, answers the below questions in an elaborate way with experi-

    mental results:

    Does the room have reflections?

    Where are the reflections inside the room?

    What is frequency response with respect to reflections in the chamber?

    What is the noise level in the room without acoustic source?

    23

  • Chapter 5. Characterization of Room Acoustics 24

    5.2 Reverberation Time

    Reverberation Time is defined as the time required (in seconds), for the average sound in

    a room to drop by 60 decibels (dB) after switching off the sound source. Reverberation

    time indicates the presence of reflections inside a room. A Room with higher Reflections

    is considered live than room with less reflections Dead.

    5.2.1 Theory

    Let us consider, we excite a sound source inside a room of Volume V . After sound in the

    room saturates, we switch off the sound source: The reverberation time (RT60) obtained

    using Sabine formula 5.1 is:

    RT60 = 0.16 VA

    (5.1)

    Reverberation Time = Constant * Volume / Absorption Area of the room

    where absorption area of the room is A = S, where is the average absorption

    coefficient of the room.

    As equation 5.1 explains reverberation depends directly on volume and indirectly

    on the area of the room. However, the above equation holds true only for the live

    room having only one absorption coefficient . However, a room having more than

    one absorption coefficient Erying [12] formula holds true to find reverberation time in

    milliseconds(ms):

    RT60 = 0.16 VS ln(1 ) (5.2)

    where,

    =1

    S

    i

    Si i (5.3)

    equation 5.3 calculates the absorption coefficient for different dimensions such as

    normal walls, walls with windows and so on as indicated by i. Hence, for our room with

  • Chapter 5. Characterization of Room Acoustics 25

    Table 5.1: Room Dimensions

    S.no Room Parameter Dimension

    1 Length 2.84 m.2 Breadth 3.06 m.3 Height 1.95 m4 Surface area 40.49 m2

    less reverberation, plugging the values of room dimensions as described in the Table 5.1

    in equation 5.3, we get theoretical reverberation time as 25 milliseconds(ms).

    To verify our theoretical calculation a verification is conducted; the following section

    will describe the brief method to measure reverberation time.

    5.2.2 Practical Measurements

    Initially, test setup was placed using ISO-3382 standard [13]. Figure.5.1 explains the

    procedure to capture RT30. Briefly, RT plots the decay of sound over the time. During

    measurement, we observed RT30 by calculating decay between -5 dB to -35 dB and

    interpolated them for RT60.

    For signal generation, we used sine burst tone as the excitation signal [14]. Sine

    burst tones is repeated sinusoidal tone with sharp attack and decay. For a room with

    reflections, the sine burst tone remains detailed. It was also easy to capture sound decay

    as compared to other sound signal like IIR (Integrated Impulse response) using balloon

    burst [15]. Further, the burst tone had good Signal to Noise Ratio(SNR) which is crucial

    for reverberation time measurement.

    For verification, we measured RT30 for 25 trails over the entire frequency sweep. This

    additional measurement gave statistical description to our experiment.

    5.2.3 Results

    As Figure.5.2 shows that reverberation time (RT30) of our acoustic cabin is 11.5 millisec-

    onds(ms). Hence, our RT60 is 23 milliseconds(ms). Also, the figure reveals a prominent

    fact that reverberation time is higher at lower frequencies (100-1000 Hz) as compared

    to the higher frequencies. Yet, it is impossible to conclude analysis based on single

    experiment.

  • Chapter 5. Characterization of Room Acoustics 26

    Figure 5.1: Flow chart explaining RT30 measurement

    To find maximum reflections and where exactly reflections happen inside the room,

    we will discuss about the Room impulse response in the following sections.

    5.3 Room Impulse Response

    5.3.1 Theory

    In signal analysis, the room impulse response is the output response shown by a system,

    when excited with an impulse input inside a room. The input impulse is characterized by

    a Dirac delta functions [16]. As we know the presence of reflections inside the chamber

    using RT30 measurement. Adding Dirac delta impulse to our room will detail out the

    reflections with respect to environment geometry. Further, reverberations are linear and

    time-invariant. Hence, the system is characterized by the given impulse response.

    Mathematically, the output of this system, y(t), can be described as a convolution

    between the systems input signal, x(t), and the room impulse response, h(t).

  • Chapter 5. Characterization of Room Acoustics 27

    Figure 5.2: RT30 plot inside acoustic chamber

    y(t) = (x h)(t) (5.4)

    Output Response = Input Signal * System Response

    Figure 5.3: Ideal RIR plot

  • Chapter 5. Characterization of Room Acoustics 28

    Using the above equation 5.4, the impulse can be calculated. Ideally, the impulse re-

    sponse plot should like in the Figure.5.3 [17]. The below paragraph explains the exper-

    imental procedure to measure RIR.

    5.3.2 Practical Measurements

    Figure.5.4 explains the logical procedure where Dirac signal is defined as :

    (x) =

    +1, x = 00, x 6= 0 (5.5)From the above equation 5.5, it is clear that signal is defined at x = 0 and zero

    elsewhere. Using this definition a Dirac wave file(.wav) is created as arbitrary function

    in the measurement instrument with one unique value and zero elsewhere.

    Figure 5.4: Flow chart explaining RIR measurement

  • Chapter 5. Characterization of Room Acoustics 29

    5.3.3 Results

    Figure.5.5 illustrates that the output response is not direct input impulse rather a re-

    sponse with variations indicating the presence of the reverberations at certain positions.

    For example, at 8 milliseconds(ms) there is a maximal peak followed by lesser peaks.

    From this, we can find the distance as:

    Speed of Sound = Distance traveled by a sound wave / Time taken for sound travel

    Using the above definition, the location of reflections is found. As an example, the

    maximal peak 8ms corresponds to glassed mirror at 2.3 meters. Similarly, for time 5.5

    ms, the distance of reflections were found to be at 1.87 meters (roof) and other smaller

    reflections came from wall behind microphone.

    Based on this measurement, we found the reflections happening at certain positions.

    Yet, we could not figure out the actual cause for this reflections and what is the possibility

    for the distortions inside the chamber. Hence, in the following section we will explore

    more about the frequency response and possible cause for acoustic distortions.

    Figure 5.5: RIR inside the acoustic chamber

  • Chapter 5. Characterization of Room Acoustics 30

    5.4 Room Frequency Response Analysis

    Frequency response analysis is a method to compute output response from a system

    with respect to a steady state oscillatory excitation. The steady state excitation is a

    sinusoidal wave. In frequency analysis the excitation is explicitly defined in frequency

    domain. The main objective to measure frequency response is to reproduce the input

    signal without any distortions. One feasible way to determine frequency response is to

    feed a swept sine wave into an acoustic system and measure the output response [18].

    The measurement procedure is same as previous sections. The standard test setup

    is observed as illustrated in the below Figure.5.6 and the test signal is sent and output

    response is measured.

    Figure 5.6: Room dimension for frequency response measurement

    The Figure.5.7 illustrates the output frequency response from the system. It is clear

    that the frequency response is not flat and has distortions. Few possibilities for this

    distortions such as:

    Due to Reflections in the chamber: Based on the exploratory measurements onreflections inside the room, the results from RT30 and RIR confirm the possibility

    for reverberations inside the chamber. This effects are produced from the walls in

    the acoustic cabin which produces unfavorable effects on the frequency response.

  • Chapter 5. Characterization of Room Acoustics 31

    Room modes at lower frequencies: Another possible cause is where the soundwaves reflected between the walls in the room and interfere with each other. This

    phenomenon is called resonance or room modes, as discussed in Section 5.5. From

    theory, every room has got room modes at certain frequencies, where a standing

    wave appears [19].

    Loudspeaker and audio amplifiers characteristics: The characteristics correspond-ing to the instruments used to produce sound inside the chamber, as discussed in

    Section 5.6.

    Figure 5.7: Frequency response measurement inside acoustic chamber

    5.5 Room Mode Calculation

    Room Modes is an important acoustical phenomenon, which can have a significant im-

    pact on the rooms characteristics. In general room modes are modal frequencies (so

    called eigenfrequencies) of a room with any shape. Hence any room has got room modes

    at certain frequencies, where a standing wave appears. Standing waves are caused by

    a perfect constructive interferences of sound waves, which are traveling between two or

    more room boundaries [20]. The standing waves appear for several modal frequencies of

    a room and can be derived with the basic wave equation.

  • Chapter 5. Characterization of Room Acoustics 32

    fres =c

    2

    (nxL

    )+(nyB

    )+(nzH

    )(5.6)

    Where: f = frequency of the mode in Hz, c = speed of sound 343 m/s, nx = numbers

    of natural oscillations (room length) (1, 2, 3, ...), ny = numbers of natural oscillations

    (room width) (1, 2, 3, ...), nz = numbers of natural oscillations (room height) (1, 2, 3,

    ...), L,B,H = length, width and height of the room in meters

    A standing wave in the 1st order of the fundamental frequency = f1 occurs, when

    half the wavelength of the excitation frequency fits between the sonically hard boundary

    surfaces.

    We can calculate standing waves of higher order from the integer multiples of the

    1st order mode as:

    f2 = 2 f1; f3 = 2 f2; f4 = 2 f3 (5.7)

    The complete table is generated for all room modes corresponding to our room using

    equation 5.6 and 5.7 :

    Table 5.2: Room Mode Calculation for Cabin

    Length=2.84 m Breadth=3.06 m Height=1.95 m

    60.24 56.8 88.31121.27 112.55 176.62181.91 168.83 264.93242.54 225.10 353.24303.18 281.38 441.55

    Graphically Figure.5.8 illustrates the distribution of the standing waves sound inside

    the room. A considerable number of isolated room modes is present for low frequencies,

    which might have an enormous impact on room characteristics. Hence, this factor is

    taken into consideration to avoid them effectively.

  • Chapter 5. Characterization of Room Acoustics 33

    Figure 5.8: Room modes inside acoustic chamber

    5.6 Room Noise Analysis

    5.6.1 Theory

    The sound which is not desired is termed as Noise. The main purpose to measure noise

    inside the room is to determine its influence on the room response. To understand them

    better, let us model the noise function using system theory.

    Figure 5.9: Noise explanation with system theory

  • Chapter 5. Characterization of Room Acoustics 34

    The model in Figure.5.9 illustrates the situation where the measured signal b(t) includes

    not only the signal v(t), the output signal from the system in response to the a(t),

    but also some additive uncorrelated noise n(t). In theory n(t) might also include some

    components transmitted from the system, but stemming from sources other than s(t).

    Using this model, we decided to plot the actual noise present inside the room.

    5.6.2 Results

    The noise plot is shown in the Figure.5.10.

    Figure 5.10: Noise inside the acoustic chamber

    To find the frequency spectrum in which the entire noise is distributed, a suitable

    frequency sweep is made with no input to loudspeaker. The above diagram clearly

    indicates the noise level is more than 10 dB at 200 Hz and goes up to 15 dB at 300-

    400 Hz frequency. This proves the presence of distortions in our room especially in

    low frequency region. Although, this analysis indicated the noise yet modeling noise to

    implement in equalization algorithm is beyond the scope the thesis. Hence, noise effect

    is used only for analysis than compensation.

  • Chapter 5. Characterization of Room Acoustics 35

    5.7 Evaluation of Room Characteristics

    Acoustics quality evaluation broadly is the area of audio engineering which includes

    various facets of evaluation from the sound perception to compression algorithms. In

    general, quality is the measure of distance between the character of an entity under

    study and the character of a target associated with this entity.

    Given the fundamental definition, we took the acoustic characteristics in a room

    as the basic entity and compared with our target dead room present in the Saarland

    University to measure quality. Although, the dimensions of the dead room is larger

    compared to the acoustic cabin in Cetecom. We were interested to evaluate software

    quality and metrics to know how far our measurement room is inadequate.

    The aforementioned experiments (Reverberation Time, Room Impulse Response,

    Loudspeaker Characteristics) were repeated with the same procedure as mentioned in

    the previous paragraphs. Let us interpret the results :

    5.7.1 Reverberations and Impulse Response

    Figure 5.11: RT-30 plot inside the dead room

    On comparing the results of Figure.5.2 and Figure.5.11 the following points are observed:

  • Chapter 5. Characterization of Room Acoustics 36

    Figure 5.12: RIR plot inside the dead room

    Reverberation time for the dead room is very less (in order of 5 ms) as comparedto actual room (around 22 ms).

    There is no evidence of reflections inside the room either in lower frequency orhigher frequency.

    Hence, from this observations we conclude that our room had reflections.

    As there is less reflections present inside the dead room then possibility of finding

    reflections is less useful. However, we find less steep curves in Figure.5.12 as compared

    to Figure.5.5.

    5.7.2 Loudspeaker Evaluation

    From the previous chapters, the frequency response is the single most important aspect

    of the performance of any audio device. Basically, it is interesting to consider that for

    as long as anyone in acoustics can remember, all electronic devices had basically flat

    frequency responses. Yet, no manufacturer of an amplifying device, would momentarily

    consider a frequency response specification from some very low frequency to some very

    high frequency.

  • Chapter 5. Characterization of Room Acoustics 37

    Yet, when we come to loudspeakers, it is important to see tolerance of 3 dB or more

    are considered acceptable. Hence, we need a mechanism to test the standard deviation

    of the loudspeaker for its tolerance.

    Hence the following Figure.5.13 for frequency response and standard deviation Fig-

    ure.5.14

    Figure 5.13: Frequency response plot inside dead room

    The above frequency response plot shows that the response is flat especially at the

    lower frequencies as compared to Figure.5.7. This fact reinforces the problem our room

    in lower frequencies where the deviation is in the loudspeakers in the higher frequencies

    which can be equalized.

    Also, from the Figure 5.14 the deviation from the loudspeaker is within the range

    of our tolerance of +/ 3 dB. Yet, at certain frequencies the deviation is high (closeto 3 dB) even in lower frequencies with respect to loudspeaker inside dead room. This

    additional information makes us to examine the possibility of inadequate quality of

    loudspeaker. Furthermore, this analysis and evaluation will be addressed for effective

    design of equalization algorithm.

  • Chapter 5. Characterization of Room Acoustics 38

    Figure 5.14: Standard Deviation plot for KS Digital loudspeakers

    5.8 Proposal for Improvements

    5.8.1 Adding Absorbers

    From the characterization of the acoustic chamber, one feasible proposal to improve the

    bad room character is to add the acoustic absorbers. As stated in the equation 5.1

    adding the absorbers is relatively straight forward solution for a room with constant

    volume. Hence, the following results stems from adding absorbers in the room.

    From Figure.5.15, the absorbers at added at the distinct positions and then room

    response is captured in the Figure.5.16. It is clear that absorbers play a crucial role in

    reducing the reflections happening inside the room. So without any algorithm compen-

    sation this improvement shows a positive sign for room improvement.

    Even the reverberation time got reduced (overall) to 8 ms as pictured in the Fig-

    ure.5.17. Yet, the response with distortions remains same in the lower frequencies.

    Adding few absorbers shows a significant improvement in reducing reflections but thats

    a one side of the medal the other side is room modes which cant be reduced by adding

    absorbers. Yet, suitable best position can be found for best effort equalization of the

  • Chapter 5. Characterization of Room Acoustics 39

    Figure 5.15: RIR: room without absorbers

    Figure 5.16: RIR: Room improvement with absorbers

  • Chapter 5. Characterization of Room Acoustics 40

    bad room acoustics. Although, complete elimination of room modes is infeasible for this

    thesis.

    Figure 5.17: RT-30: Room improvement with absorbers

    5.9 Summary

    To summarize:

    Room acoustic analysis consists of three experiments namely Reverberation Time(RT), Room Impulse Response (RIR) and Room frequency response or Loud-

    speaker frequency response. In these analysis, distortions present in the non-linear

    room frequency response of the acoustic chamber are due to reflections as con-

    firmed by RT and RIR measurements.

    Further, to detail why distortions are observed apart from reflections: Room modecalculations and Room noise analysis are experimented.

    Based on the Room acoustic analysis, suitable acoustic absorbers are placed withinthe cabin and it significantly reduced reflections. Yet, the other probable distor-

    tions caused by either Room modes or Room noise remains unaltered. For all

    intents and purposes, we avoid Room modes and Room noise by finding optimal

    positions to place loudspeaker and microphone inside the chamber.

  • Chapter 5. Characterization of Room Acoustics 41

    Hence, the characterization of room acoustics builds suitable notion on how thecabin behaves acoustically, based on the observed characters a suitable equalization

    algorithm will be designed in the next chapter.

  • Chapter 6

    Equalization

    The present chapter explains the idea and implementation of equalization algorithm

    with integration and verification to the system. The approach briefly describes:

    Equalization Design and Implementation: An equalization algorithm is designedand implemented in a way that the performance of the algorithm is guaranteed

    for its efficiency and accuracy. A suitable position is found for the best effort

    performance of the algorithm considering bad room effects.

    Equalization Verification: Verification explains the statistical accuracy and algo-rithm run time to guarantee best results. Here, we also discuss the behavior of

    algorithm in different cases with correlation to room analysis conducted.

    6.1 Equalization Design and Working

    This section focuses on the architecture of equalization algorithm in specific which will

    be the novelty of this thesis.

    6.1.1 Programming Decision

    To develop an accurate and efficient equalization algorithm, we asked few questions to

    derive our algorithm like:

    42

  • Chapter 6. Equalization 43

    How much accuracy can be achieved? In section 5.4, we attempted to have stan-dard test setup to find best positions in the room to avoid reflections. Hence, to

    achieve good accuracy we will develop an algorithm for one best position based on

    several trails.

    How to make the algorithm efficient with minimal run time? We saw in Section4.1.1, that the SCPI commands are fast enough to communicate with the instru-

    ment. Therefore, in order to achieve efficiency, SCPI commands with COM library

    will be used to develop the algorithm to have minimal run time.

    With this reasoning, we chose to develop the algorithm using these concepts in close

    association.

    6.1.2 Theory and Block Diagram

    The main goal of the algorithm is to reduce bad room characteristics with accuracy of

    +/ 0.5 dB and be time efficient(

  • Chapter 6. Equalization 44

    Figure 6.1: Equalization algorithm design: adaptation from M.Sc seminar

    say 90 dB or any value of interest. The entire function completes the calibration

    with an accuracy of +/ 0.02 dB. The new voltage is found using equation 6.2.

    To find initial voltage say xdBV correction at 1k Hz for a target SPL of 90 dB

    SPL(-43 dBV) the following relationship is used :

    xdBV = 20 log10(V/V0) (6.1)

    Where V is the correction voltage and V0 is reference voltage normally 1 V.

    Rearranging the above equation yields,

    V = 10xdBV20 (6.2)

    The above equation 6.2 is run iteratively to find the actual voltage. For our

    example at 90dB SPL the applied voltage is 69.7 mV(millivolt) which is to be

    applied to all frequencies is filled into column vector:

    Vnormal = V I (6.3)

  • Chapter 6. Equalization 45

    here I is identity matrix

    Vnormal =

    69.7

    69.7

    ..

    ..

    69.7

    (6.4)

    Sweep and Store Values: Based on applied correction voltage from equation6.3, a frequency sweep Freq is made and the output response is:

    Freq =

    200

    205.67

    ..

    ..

    10, 000

    (6.5)

    Xuneq =

    41.3841.39..

    ..

    33.24

    (6.6)

    Where, Xuneq is th output in dBV and Freq is the frequency points from 200-10k

    Hz consisting of 150 data points.

    First Equalization: Based on the frequency sweep at the defined target soundpressure level, the first equalization finds a offset factor as given in equation 6.7.

    To find offset between the target SPL and measured value (Xuneq), we derive an

    offset factor KFactor as using below:

    KFactor(i) 10targetSPLXuneq(i)

    20 (6.7)

    For our first iteration, target SPL = -43.0 and Xuneq = -41.38 (for 200 Hz) on

    plugging in the equation 6.7 we get:

  • Chapter 6. Equalization 46

    KFactor =

    0.7784

    0.8114

    ..

    ..

    0.3175

    (6.8)

    So, first equalized value Eequal is generated as:

    EEqual = Vnormal KFactor (6.9)

    The above equation produces right compensation for the bad room acoustics dur-

    ing equalization. So, the first equalization means applying the equation 6.9 and

    equalizing the response from the system.

    Check Accuracy and Second Equalization: Further, we check the accuracyand if its not as desired (+/ 0.5 dB) then voltage is further rounded off for moreprecision and algorithm runs in second iteration as termed as second equalization

    System Halt: Finally, after second equalization, system is switched off. Theobtained data is processed and values are stored.

    A self-explanatory flowchart is illustrated in Figure.6.2

    6.2 Setup and Results

    Figure.6.3 shows the best case arrangement for the the room before equalization to avoid

    room modes and acoustic noise inside the room. Here, this position is found in way that

    it is close to (2/3)rd away from the mirrored wall in breadth. It is also (1/2)th in the

    length. Based on several trail and error methods this position is found. Hence, on this

    position algorithm works stably with guaranteed accuracy.

    Figure.6.4 shows the output equalized response from the system with respect to 90

    and 80 dB SPL. As we see from the figure the whole response is equalized at target SPL.

    The accuracy of the overall algorithm with respect to 90 and 80 dB is +/ 0.2 dB forall frequencies. The another important fact is that the output response is not spurious

    and is close to a straight line at the target SPL.

  • Chapter 6. Equalization 47

    Figure 6.2: Equalization module flow chart

  • Chapter 6. Equalization 48

    Figure 6.3: Best case setup inside acoustic chamber

    For the target SPL of 70 and 60 dB SPL the output is indicated in the Figure.6.5.

    For 70 dB SPL the outputs behaves similar to 90 and 80 dB SPL. Yet, for 60 dB SPL

    the deviation is up to than +/0.2 dB. Also, the curves gets little spurious in the lowerfrequency region. However, the curve looks straight overall with satisfying accuracy.

    From the Figure.6.6, the curve is not straight or ideal as per the requirement. How-

    ever, the output is equalized at the target SPL which is 50 dB SPL and is within the

    range of +/ 0.5 dB SPL.

    Main reasons for not getting ideal output response curve is :

    Influence of noise level at lower sound pressure levels

    Room Modes at lower frequencies

    The above two mentioned points were already concluded in the Chapter 5. However, we

    will do detailed statistical analysis in the below sections.

  • Chapter 6. Equalization 49

    Figure 6.4: Equalization results for 90 and 80 dB SPL

    6.3 Verification

    To verify the accuracy of our algorithm, we conducted two experiments with a goal to

    test the algorithm:

    6.3.1 Verification at 50 dB SPL

    Here the algorithm is tested at 50 dB SPL where the same correction factor as given in

    equation.6.8 is applied in two trails without changing or modifying values. As per the

    definition of testing, we should observe a repeatable results again. Yet, the results are not

    same which reasons the inadequacy of the algorithm at 50 dB SPL. As stated, presence

    of room modes at lower frequencies makes the system immune to applied voltage. Also,

  • Chapter 6. Equalization 50

    Figure 6.5: Equalization results for 70 and 60 dB SPL

    Figure 6.6: Equalization results for 50 dB SPL

  • Chapter 6. Equalization 51

    the points where the deviation is higher than 0.5 dB are similar points or frequencies as

    observed in the noise response from the room around 200-400 Hz illustrated in Figure.6.7.

    Although, the verification between two trails is not 100 % accurate, yet our algorithm

    does better by leaving 1% error out of 150 data points which is still acceptable.

    Figure 6.7: Verification at 50 dB SPL

    6.3.2 Statistical Accuracy at 50 dB SPL

    To find the statistical accuracies at the 50 dB SPL, the equalization algorithm is executed

    several trails(approx.25) to observe the deviation in the equalized points as shown in

    Figure.6.8. Every time the correction voltage is changed and same voltage is not applied

    again. This experimental procedure gave results conforming the deviation in accuracy

    up to +/ 0.5 dB specially at lower frequencies(less than 1000 Hz). This experimentagain proves that our solution is still acceptable in terms of accuracy at lower sound

    pressure levels (50 dB).

  • Chapter 6. Equalization 52

    Figure 6.8: Statistical Deviation with +/ 0.5 dB

    6.4 Algorithm Runtime

    Having accurate results from the equalization algorithm is one side of the medal, another

    side is to be time efficient in equalizing the values within the defined time-budget (less

    than 1 hour). To see, how far our developed algorithm is time efficient, a comparison

    bar plot is made as illustrated in the Figure.6.9. From the diagram, the algorithm

    runs for about 10 minutes to equalize 150 data points which makes our algorithm more

    efficient. Similarly, with increase in the samples or data points the run time increases

    proportionally. Yet, in all the cases the run time is well within the time-budget. The

    run time computed here is from start of the system with user intervention and actual

    equalization completed.

    6.5 System Testing

    Here the utilization of the equalization algorithm is done by the system to evaluate

    the performance of the hearing aid algorithms using default R & S UPV software and

    implemented novel equalization algorithm.

  • Chapter 6. Equalization 53

    Figure 6.9: Algorithm run time as size of data increases

    Figure 6.10: Field of frequency response: using R &S system

    6.5.1 Comparison of Field Frequency Response

    The field of frequency response is the frequency sweep algorithm done in the hearing

    aid at various SPL. The algorithm runs from the lowest sound pressure level to the

    highest sound pressure(50 dB SPL to 90 dB SPL) spanning from 100-10k Hz. This

    complete algorithm is offered in UPV-K7 which results in the output response as shown

    in the Figure.6.10 . As plotted in the figure the algorithm does not produce accurate

    results as per the wish-list which is +/ 0.5 dB. The algorithm leaves two or manypoints deviated.

  • Chapter 6. Equalization 54

    Figure 6.11: Field of frequency response: using novel equalization algorithm

    To avoid this problem, our novel equalization algorithm is verified for the system at

    various SPL. The results are illustrated in the Figure.6.11. As we compare the results

    from default R & S UPV equalization algorithm to our new equalization algorithm, defi-

    nitely the new proposed algorithm outperforms the proprietary solutions with increased

    accuracy and with increased efficiency satisfying IEC standards.

    6.6 Summary

    To summarize:

    We developed a complete novel equalization algorithm based on our understandingof room acoustics by suitable characterization measurements inside the chamber.

    The possible distortions are compensated by finding suitable positions to placeloudspeaker and microphones inside the chamber.

    The developed algorithm satisfies the IEC standard requirements at all soundpressure levels. Yet, few points (less than 1% of all data points) may deviate more

    than +/ 0.5 dB at 50 dB SPL which is still adequate for hearing aids testing.

    System testing further verifies our eminence of our developed algorithm as it signifi-cantly outperforms the existing solutions (both R & S K-7 and Cetecom solutions).

  • Chapter 7

    Conclusion and Future Work

    The present Master thesis implements an automated remote control based solutions

    to test and measure electro-acoustical parameters in hearing aids by developing novel

    equalization algorithm to compensate unfavorable room effects. In particular, the thesis

    presents detailed analysis of the room characteristics with possible improvements to the

    room acoustics. Based on the technical analysis, a suitable equalization algorithm is

    designed and implemented so as to satisfy the performance requirements introduced

    by IEC 60118 standard for hearing aids measurements and discusses the ability of the

    developed algorithm to meet the requirements in terms of accuracy and time efficiency.

    The characterization of the room acoustics is well explained by techniques namely:

    Reverberation Time, Room Impulse Response in the acoustic chamber which indicated

    the presence of reflections inside the room. By adding few acoustic absorbers the re-

    flections inside the room got compensated. However, the distortions in the frequency

    response of the room is not completely compensated by absorbers as it is caused by

    the presence of room modes and acoustic noise at lower frequencies and lower sound

    pressure levels. Practically, we tried to avoid room modes and acoustic noise by placing

    the acoustic components optimally inside the chamber during equalization. Hence, the

    characterization shows that with suitable absorbers the room response can be improved

    significantly.

    The novel equalization algorithm developed from the room characterization with

    optimal positions shows a significant improvement in the accuracy and time efficiency

    as confirmed by system verification and statistical analysis. The performance evaluation

    55

  • Chapter 7. Conclusion and Future Work 56

    conducted between the new equalized algorithm and default proprietary solutions indi-

    cates that the new algorithm outperforms default algorithm in terms of accuracy and

    efficiency. Although, the developed solution is not completely accurate in lower sound

    pressure levels such as 50 dB. Still, our solution produces a reliable and acceptable re-

    sults in terms of accuracy +/ 0.5 dB and run time efficiency within 1 hour which formsthe basis to satisfy IEC standards for hearing aids compliance testing.

    We claim novelty in the thesis for having a stringent requirements to develop

    equalization algorithm with +/ 0.5 dB which runs within 1 hour. Throughout ourwork, we had one crucial challenge as to develop an equalization algorithm without using

    any post-processing of signals which makes our design and implementation unique. Most

    importantly, our developed solution accomplishes an acceptable and reliable results at all

    sound pressure levels satisfying strict requirements from IEC standards which elevates

    novelty by all measures.

    The thesis presents a technically feasible implementation to measure electro-acoustical

    entities of hearing aids by systematic analysis of the room acoustics by developing a

    novel self-calibrating algorithm to equalize unfavorable effects of bad room acoustics.

    The satisfaction of performance metrics such as accuracy and time efficiency set by the

    IEC standards elevates the capability of our solution for the next-generation test and

    measurement systems in the context of electro-acoustical measurements. Though, the

    developed solutions concentrates on hearing aids measurements yet the automated solu-

    tions constitute a vibrant research and development field especially in the area of room

    acoustics. Hence, a variety of technical challenges are therefore awaiting solutions.

    There is always room for improvement. The evaluation of the equalization algorithm

    at lower sound pressure levels and lower frequencies brings up new ideas and challenges,

    One feasible implementation involving room modal based compensation focusing on 50

    dB SPL can be realized. However, before developing modules on equalization several

    frequency response experiments with new loudspeakers can be done as a quantitative

    measures as we doubt the performance of existing loudspeaker. Similarly, an effective

    characterization of the sound system amplifiers needs to be conducted as a possible

    future work.

  • Bibliography

    [1] Michael Valente. Hearing aids: Standards, options, and limitations. Thieme, 2002.

    [2] James Jerger. Behavioral correlates of hearing-aid performance. Bull. Prosthetics

    Res. Spring, 1967.

    [3] John N Mourjopoulos. Digital equalization of room acoustics. Journal of the Audio

    Engineering Society, 42(11):884900, 1994.

    [4] Nikolai Bisgaard, Marcel SMG Vlaming, and Martin Dahlquist. Standard audio-

    grams for the iec 60118-15 measurement procedure. Trends in amplification, 14(2):

    113120, 2010.

    [5] Matti Karjalainen, Esa Piirila, Antti Jarvinen, and Jyri Huopaniemi. Comparison

    of loudspeaker equalization methods based on dsp techniques. Journal of the Audio

    Engineering Society, 47(1/2):1431, 1999.

    [6] Aki Makivirta, Poju Antsalo, Matti Karjalainen, and Vesa Valimaki. Modal equal-

    ization of loudspeaker-room responses at low frequencies. Journal of the Audio

    Engineering Society, 51(5):324343, 2003.

    [7] Arie JM Kaizer. Modeling of the nonlinear response of an electrodynamic loud-

    speaker by a volterra series expansion. Journal of the Audio Engineering Society,

    35(6):421433, 1987.

    [8] Wolfgang Klippel and Robert Werner. Loudspeaker distortionmeasurement and

    perception.

    [9] SCPI Consortium et al. Standard commands for programmable instruments, 1999.

    57

  • Bibliography 58

    [10] Dany Cheij. A software architecture for building interchangeable test systems. In

    AUTOTESTCON Proceedings, 2001. IEEE Systems Readiness Technology Confer-

    ence, pages 1622. IEEE, 2001.

    [11] Steve Burbeck. Applications programming in smalltalk-80 (tm): How to use model-

    view-controller (mvc). Smalltalk-80 v2, 5, 1992.

    [12] Antonio Pedro O Carvalho. The use of the sabine and eyring reverberation time

    equations to churches. The Journal of the Acoustical Society of America, 97(5):

    33193319, 1995.

    [13] John S Bradley. Using iso 3382 measures, and their extensions, to evaluate acous-

    tical conditions in concert halls. Acoustical science and technology, 26(2):170178,

    2005.

    [14] Angelo Farina. Advancements in impulse response measurements by sine sweeps.

    In Audio Engineering Society Convention 122. Audio Engineering Society, 2007.

    [15] Jukka Patynen, Brian FG Katz, and Tapio Lokki. Investigations on the balloon

    as an impulse source. The Journal of the Acoustical Society of America, 129(1):

    EL27EL33, 2011.

    [16] Floyd E Toole. Loudspeakers and rooms for sound reproductiona scientific review.

    Journal of the audio engineering society, 54(6):451476, 2006.

    [17] Heinrich Kuttruff. Room acoustics. CRC Press, 2009.

    [18] Swen Muller and Paulo Massarani. Transfer-function measurement with sweeps.

    Journal of the Audio Engineering Society, 49(6):443471, 2001.

    [19] Thomas D Rossing. Springer handbook of acoustics. Springer, 2007.

    [20] Oscar J Bonello. A new criterion for the distribution of normal room modes. Journal

    of the Audio Engineering Society, 29(9):597606, 1981.

    AcknowledgementsAbstractList of FiguresList of TablesAbbreviations1 Introduction1.1 Background1.2 Motivation1.3 Outline

    2 Problem Formulation2.1 Need for Calibration2.2 Equalization by R&S K-7 Software2.3 Equalization by Cetecom software2.4 Research Questions2.5 Plan of Attack

    3 Acoustics: Overview3.1 Fundamentals of Sound3.2 Propagation of Sound3.3 Sound Level and the Decibel3.4 Acoustic Components3.5 Room Acoustics3.6 Absorption on Surfaces3.7 Early Reflections

    4 Methodology4.1 Software Design and Architecture4.1.1 SCPI: Overview4.1.2 VISA Library4.1.3 Event Driven Programming4.1.4 MVC: Design Pattern

    5 Characterization of Room Acoustics5.1 Room Acoustic Analysis5.2 Reverberation Time5.2.1 Theory5.2.2 Practical Measurements5.2.3 Results

    5.3 Room Impulse Response5.3.1 Theory5.3.2 Practical Measurements5.3.3 Results

    5.4 Room Frequency Response Analysis5.5 Room Mode Calculation5.6 Room Noise Analysis5.6.1 Theory5.6.2 Results

    5.7 Evaluation of Room Characteristics5.7.1 Reverberations and Impulse Response5.7.2 Loudspeaker Evaluation

    5.8 Proposal for Improvements5.8.1 Adding Absorbers

    5.9 Summary

    6 Equalization6.1 Equalization Design and Working6.1.1 Programming Decision6.1.2 Theory and Block Diagram

    6.2 Setup and Results6.3 Verification6.3.1 Verification at 50 dB SPL6.3.2 Statistical Accuracy at 50 dB SPL

    6.4 Algorithm Runtime6.5 System Testing6.5.1 Comparison of Field Frequency Response

    6.6 Summary

    7 Conclusion and Future WorkBibliography