University of Southern California Dissertations and Theses

An open ear canal sound delivery system

(USC Thesis Other)

An open ear canal sound delivery system

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content AN OPEN EAR CANAL SOUND DELIVERY SYSTEM Copyright 2004 by Neeta Chandwani A Thesis Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVESITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements of the Degree MASTER OF SCIENCE (BIOMEDICAL ENGINEERING) August 2004 Neeta Chandwani Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. U M I Number: 1422387 INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. ® UMI UMI Microform 1422387 Copyright 2004 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ii Acknowledgements I take this opportunity to thank my advisor Dr. Sigfrid Soli, at House Ear Institute for guiding me in this research work. With his enthusiasm, his inspiration, and his great efforts to explain things clearly and simply, he made this project truly a fine learning experience. I would also like to graciously thank Dr. Michael Khoo, Chair Biomedical Engineering Department, USC and Dr. Manbir Singh, Professor, Biomedical Engineering Department, USC for their help in development of this thesis. I am very grateful to my coworkers at the House Ear Institute, Dan Freed, Andy Vermiglio and Peggy Nasoordeen for providing a comfortable atmosphere to work and with their endless help in the laboratory work. I would also like to thank Buzz Moran and David Krubsack from Implanted Acoustics Inc. for their constant support both financially and technically during the entire project. Lastly, I am thankful to my family and friends for their support. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iii Table of Contents Acknowledgments...................................................................................................................ii List of Tables........................................................................................................................... v List of Figures......................................................................................................................... vi Abstract....................................................................................................................................ix 1.0 Introduction 1.1 Human Auditory System...................................................................................1 1.2 Hearing Loss......................................................................................................3 1.3 Auditory Prosthesis...........................................................................................4 1.3.1 Hearing Aids........................................................................................4 1.3.2 Implanted Auditory Prosthesis........................................................... 7 1.4 Clinical Significance.........................................................................................9 2.0 System Design and Measurement Methodologies 2.1 Objective.......................................................................................................... 11 2.2 System Configuration.....................................................................................14 2.3 Measurement and Analysis Methods 2.3.1 Model Design using FIR filters........................................................ 18 2.3.2 Estimating Impulse Response of the system.....................................19 2.3.3 Denconvolution T echnique.............................................................. 20 2.3.4 Estimation of feedback.....................................................................21 2.4 Signal Processing............................................................................................23 3.0 Research Design and Implementation 3.1 Scope of Thesis............................................................................................... 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2 Instrumentation...............................................................................................25 3.3 Azimuthal Dependence of head related transfer functions......................... 30 3.4 Calibrations..................................................................................................... 31 3.5 Acoustic Transfer Functions..........................................................................33 3.6 Summation Results.........................................................................................42 3.7 System Limitation...........................................................................................46 4.0 Model Validation 4.1 Hearing Aid Fitting Rules.............................................................................. 48 4.2 V alidation Results...........................................................................................51 4.2.1 Front...................................................................................................51 4.2.2 Ipislateral........................................................................................... 52 4.2.3 Contralateral...................................................................................... 54 5.0 Conclusion 5.1 Conclusion...................................................................................................... 56 5.2 Further Work................................................................................................... 56 References...............................................................................................................................58 Appendix.................................................................................................................................61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V List of Tables 1. Receiver S aturation Levels.......................................................................................17 2. Reference Conditions for Validation.......................................................................49 3. FG Series Microphone Specifications.....................................................................64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. List of Figures 1. The Human Ear.............................................................................................................2 2. Behind-the-ear Hearing Aid.........................................................................................4 3. In-the-ear Hearing Aid................................................................................................. 6 4. Completely-in the ear Hearing Aid............................................................................. 6 5. In-the canal Hearing Aid..............................................................................................6 6. Cochlear Implant...........................................................................................................7 7. BAHA............................................................................................................................8 8. Middle ear implant....................................................................................................... 9 9. Prototype Sound Delivery System............................................................................ 11 10. System Overview........................................................................................................14 11. DHP 100 EVM Block Diagram Overview...............................................................16 12. Estimating Impulse Response of any system........................................................... 20 13. Deconvolution Technique..........................................................................................21 14. B&K type 2690 Sound Level Meter.........................................................................26 15. FG 3329 Microphone................................................................................................. 27 16. TI DHP 100 Board..................................................................................................... 27 17. Knowles Cl 8409 Receiver........................................................................................28 18. Block Diagram of Signal Processing Software........................................................ 29 19. Azimuthal dependence of HRTF.............................................................................. 31 20. Calibrations.................................................................................................................31 21. Delays in four Azimuths............................................................................................33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vii 22. Laboratory set up for ear canal measurements........................................................ 34 23. Impulse responses for ear canal in four directions...................................................35 24. Front and Back Transfer Functions........................................................................... 36 25. Ipsilateral and Contralateral Transfer Function....................................................... 37 26. Laboratory set up for FG 3329 Response Measurements....................................... 38 27. Source-to-microphone response for 4 directions.....................................................39 28. Laboratory set up for measuring response of TI DHP 100..................................... 40 29. TI DHP 100 Response...............................................................................................40 30. Cl 8409 Receiver Response.......................................................................................41 31. Reference Microphone response............................................................................... 42 32. Laboratory set up for measuring combined response..............................................43 33. Front Azimuth Summation Results...........................................................................43 34. Back Azimuth Summation Results...........................................................................44 35. Ipsilateral Azimuth Summation Results...................................................................45 36. Contralateral Azimuth Summation Results............................................................. 46 37. Percentage THD......................................................................................................... 47 38. No Harmonics at 103dB.............................................................................................38 39. Harmonics at 104dB.................................................................................................. 38 40. GUI for Interface Software........................................................................................50 41. Front Azimuth with FBC On.....................................................................................52 42. Front Azimuth with FBC Off................................................................................... 52 43. Ipsilateral Azimuth with FBC On............................................................................. 53 44. Ipsilateral Azimuth with FBC Off..........................................................................54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. viii 45. Contralateral Azimuth with FBC On........................................................................55 46. Contralateral Azimuth with FBC Off.......................................................................56 47. Weiner Filter Program............................................................................................... 62 48. HTD............................................................................................................................ 63 49. DSC.exe...................................................................................................................... 63 50. Simulink Diagram...................................................................................................... 64 51. TI DHP 100 Simulink Model.................................................................................... 67 52. FG 3329 Response in Front and Back Direction.....................................................67 53. FG 3329 Response in Contralateral and Ipsilateral Direction................................68 54. FG 3229 response as from data sheet.......................................................................68 55. Cl 8409 Performance Specifications........................................................................69 56. Receiver Response..................................................................................................... 69 57. Receiver Saturation Levels........................................................................................69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ix Abstract We are developing a novel hearing aid sound delivery system capable of producing adequate sound pressure levels in the ear canal of patients with moderate and severe hearing loss while keeping the ear canal open at all times. This thesis establishes the development and working of an electro-acoustic computer model of this sound delivery system. Such a model would help in predicting the performance of this complex electro acoustic system and will help in understanding the dependence of its performance on acoustic, electro acoustic and physical parameters in a quick and efficient way. The essential components include a Knowles Electronics microphone (FG 3329); a Texas Instrument Digital Hearing Processor (DHP 100) and a Knowles Electronics Cl 8409 receiver along with plastic tubing partially inserted into the subject’s ear canal. This system is modular and scalable to include effects of non-linearties associated with noise floor and harmonic distortion. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 Chapter 1 Introduction The human auditory system is a very sophisticated and complicated signal processing system that is capable of perceiving and analyzing very complex sounds and discriminating subtle changes in sound. These characteristics are crucial for the perception and recognition of speech and for interpretation of the sound patterns encountered in daily life, as well as for listening to music. The normal hearing system can be damaged due to aging, otologic diseases, exposure to loud noises, ototoxic drugs and other reasons. This will often result in a communication handicap, due to loss of sensitivity and impaired discrimination of speech sounds as well as other auditory stimuli. Due to a growing urbanization trends hearing loss has continued to become a very common problem. Approximately 28 million people in the United States suffer from hearing loss. Several authors have demonstrated that people with hearing loss suffer significant emotional, social, and communication dysfunction. Other studies have shown that people who use a suitable auditory prosthesis experience a significant improvement in social and emotional function, communication and cognitive function, and depression over those who do not. The dynamic range of the human auditory system, which is the interval between the softest and loudest sounds that the ear can hear, is more than 120 decibels. Frequencies between 20 Hz and 20000 Hz are audible to the human ear. The auditory canal can resonate and amplify sounds within a frequency range of about 2000 Hz to 5500 Hz by up to a factor of lO.This thesis presents a non-linear auditory model of a novel sound delivery hearing prosthesis capable of a dynamic range of about 75 dB and gain of 30dB. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 1.1 Human Auditory System Sgmtcarcdor 'Ccmak Auditory " N&rw C ochlm Round Window • s _ Eustachian Tube Figure 1: The Human Ear, http://www.audiohearing.co.uk The ear is made up of three areas: the outer, middle, and inner ear. The outer ear is very important for collecting sound waves. It is made up of the pinna and the ear canal. The pinna, the actual physical outward appearance of the ear, receives sound waves and begins to funnel them into the ear canal. The ear canal is also known as the auditory meatus, which is basically a convoluted tube. The next part of the ear, the tympanic membrane, is the beginning of the middle ear. The eardrum is crucial in the ability to hear. The tympanic membrane leads to a chain of small bones known as the malleus (hammer), incus (anvil), and the stapes (stirrup). The stapes is ended with the footplate, a bone that looks like a stirrup. This area is known as the middle ear or the tympanic cavity. Located at the bottom of this area is the Eustachian tube, which leads down to the throat. Its main purpose is to maintain the equalization of pressure between the tympanic cavity and the atmosphere as the cells of its surface absorb the air in the cavity. The next area is the inner ear. This area contains many important structures to the hearing process. It begins with the oval window, which is struck by the footplate of the Stapes. The C M W uxto* Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 cochlea is the area where most sound is transmitted from waves into impulses. Within the cochlea there are two different types of hair cells. One or two radial fibers innervate the inner hair cells and each radial fiber is attached to one or two hair cells whereas the external hair cells have many innervations. This movement swaying across the different hair cells creates impulses that are sent to the brain through the eighth cranial nerve. It would now occur to one about the consequences of breakdown of such a marvel of natural engineering. Deformities in the auditory system result in hearing loss and in severe cases in deafness. Next section talks briefly on this. 1.2 Hearing Loss Hearing loss and deafness affects individuals of all ages and may occur at any time from infancy through old age. Impairments in hearing can occur in either or both ears. Hearing loss is generally described as slight, mild, moderate, severe, or profound, depending upon how well a person can hear the intensities or frequencies most greatly associated with speech. In general, hearing loss is categorized into three broad categories: Conductive - Conductive hearing loss occurs when sound is not conducted efficiently through the outer and middle ears, including the ear canal, eardrum, and the tiny bones, or ossicles, of the middle ear. Conductive hearing loss usually involves a reduction in sound level, or the ability to hear faint sounds. This type of hearing loss can often be corrected through medicine or surgery. Sensorineural - Sensorineural hearing loss occurs when there is damage to the inner ear (cochlea) or to the nerve pathways from the inner ear to the brain. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 Mixed - Mixed hearing loss is a combination of both conductive and sensorineural hearing loss occurring at the same time. Both the middle and inner ear are involved. With mixed hearing loss, the conductive part may be treated, but the sensorineural part is usually permanent. 1.3 Auditory Prosthesis Field Hearing instruments can provide significant help for most hearing losses although they cannot restore hearing to normal. Few of the hearing impairments common among people respond to medical or surgical treatment. However, a variety of approaches, such as the use of hearing aids, other assistive listening devices, and aural rehabilitation techniques, can be used to compensate for hearing impairment. Brief overview of commonly available hearing prosthesis is presented below: 1.3.1 Hearing Aids: A huge variety of hearing aids have been available for over hundred years now. Their function is to amplify sound to such a level that hearing-impaired people can both detect and most importantly make effective use of acoustic signal. All air conduction hearing aids have the same basic components, which include a microphone that transduces sound to electrical energy, and amplification stage, an output transducer called a receiver, and a battery to power the electronics. These instruments are essentially sound amplifiers. No matter what the size, style or manufacturer, all hearing aids have the same basic components. However hearing aids only partially overcome the deficit associated with a hearing loss. An estimate of 4.5 million Americans use hearing aids and of those that own hearing aids 12% report never using them and of those that do wear them only 58% report being fully satisfied. Nevertheless hearing aid remains the most common auditory prosthesis. Some types of conventional hearing aids are described here: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 Behind-the-ear (BTE) hearing aids fit comfortably behind-the-ear and are attached to a soft custom ear mold. Sound is routed through the ear mold into the ear. The microphone is located at the top of the hearing aid near the ear hook. The battery, amplifier and receiver are all inside the case that fits behind the ear. Figure 2: Behind-The-Ear Hearing Aid, http://www.hearingaidhelp.com/ In-the-ear (ITE) hearing aids can be used for a wide range of hearing losses. Due to their size, ITE hearing aids allow for larger sound amplifiers and more features such as a telephone switch. The hearing aid case is custom made out of a hard plastic material. As shown, the hearing aid case houses all of the miniaturized hearing aid parts. Figure 3: In-The-Ear Hearing Aid, http://www.hearingaidhelp.com/ Completely-in-the-canal (CIC) hearing aids are the smallest size of hearing aids, practically invisible to an observer. Custom designed to fit the wearer's ear; CIC hearing aids fit deep inside the ear canal and most closely resemble the natural hearing. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 4: Completely-in-the-canal Hearing Aid, http://www.hearingaidhelp.com/ In-the-canal (ITC) hearing aids fit into the ear canal. They are only slightly larger than the completely-in-the-canal (CIC) hearing aid but smaller than the in-the-ear (ITE) hearing aid. Figure 5: In-the-canal Hearing Aid, http://www.hearingaidhelp.com/ 1.3.2 Implanted Auditory Prostheses Cochlear Implants (Cl): Cochlear implants compensate for damaged or non-working parts of the inner ear. It electronically finds useful sounds and then sends them to the brain. A cochlear implant electronically stimulates the auditory nerve, in contrast to the usual process of acoustical stimulation. In short, sound is picked up by the microphone and sent by a cord to the speech processor, where the sound is encoded. The signal then is sent back by the cord to the transmitter. Next, the transmitter sends the signal as a radio signal to the receiver/stimulator. The receiver/stimulator sends the signal by an internal Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 wire to the electrode array. Finally, the array stimulates the auditory nerve. Hearing through an implant may sound different from normal hearing, but it allows many people to communicate fully with oral communication in person and over the phone. Figure 6: Cochlear Implant, http://www.cochlearamericas.com Auditory Brain Stem Implants (ABI): Surgical removal of the tumors on the auditory nerve is necessary and often requires severing or cutting the nerve. This results in total loss of hearing. These patients cannot be helped by conventional hearing aids or cochlear implants. The ABI is designed to restore a sense of sound to people who experience total hearing loss when the cranial hearing nerves are damaged through tumor removal. The device restores the ability to detect environmental cues and certain speech sounds. It does not restore hearing. An electrode array is surgically implanted into the brain stem and electrically stimulates the area that normally receives the electrical signal from the auditory nerve linked to the ear. The patient wears a pocketsize speech processor attached to a microphone that picks up sound and changes it into electrical pulses that are transmitted to the implanted electrode array. Bone Anchored Hearing Aids (BAHA): The BAHA is a surgically implantable system for treatment of hearing loss that works through direct bone conduction. The BAHA consists of three parts: a titanium implant, an external abutment, and a sound processor. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 The system works by enhancing natural bone transmission as a pathway for sound to get implant is placed during a short surgical procedure and over time naturally integrates with the skull bone. For hearing, the sound processor transmits sound vibrations through the external abutment to the titanium implant. The vibrating implant sets up vibrations within the skull and inner ear that finally stimulate the nerve fibers of the inner ear, allowing hearing. Figure 7: BAHA, http://www.yb.gmm.gu.se Middle-Ear Implants: The middle ear implant is a useful hearing instrument and is quite different from traditional hearing aids. Middle ear implants work by vibrating the middle ear bones, rather than by producing audible sound. The reported benefits of middle ear implants are elimination of the occlusion effect, elimination/reduction of feedback, reduction in distortion, improved clarity, as well as some cosmetic advantages. to the inner ear, bypassing the external auditory canal and middle ear. The titanium « Figure 8: Middle Ear Implants, http://www.healthyhearing.com Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 1.4 Clinical Significance It becomes obvious from the discussion above that there is a multitude of hearing prosthesis available for use in case of hearing impairments. In this thesis however the discussion would be primarily focused on acoustic amplification and hearing aid devices. As briefly mentioned above hearing aids remain the popular choice as auditory assistive devices but there are some factors that interfere with increased use of hearing aids. These include problems in the design or function of the aid; problems in selecting an appropriate aid for the individual; inability of the individual to adjust to the aid; and disagreement among hearing specialists about who can benefit from a hearing aid. The frequency with which each of these problems occurs is not known. Developing solutions for them is a potential area for cooperative research by the hearing aid industry and hearing specialists (physicians, audiologists, and hearing aid dealers). Some people buy hearing aids that are not well matched to their needs. Sometimes this occurs because they purchase an aid without having a complete hearing evaluation to identify their specific hearing deficits. Hearing aids generally amplify all environmental sound, including background noise. Although design modifications can improve the speech-to-noise ratio, some hearing aid users continue to have problems tuning out background noise. Apart from this a very persistent problem that hearing aid users face is due to the design of most receivers. It causes the ears to be blocked giving rise to ‘occlusion effect’. One of the first things that occur with this is the change in the sound of one’s own voice. Vibrations from the users own voice are actually trapped inside the ear canal and may sound like an echo. Own voice sounds different because user has plugged up his ears. We propose a novel sound Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 delivery technology in which the ear canal is left open and the transducer (receiver) could rest on the ear pinna like a BTE module or may be implanted inside (the idea is to keep ear canal open). This amplified sound energy is then mixed with sound delivered naturally and directly into the ear canal. This methodology is helpful in overcoming lot of negative effects as generated by hearing aids that partially or fully occlude the ear canal. It has been established during the course of this research and prior work that such a system is capable of providing high gain, low distortion in outputs and definitely a more aesthetically acceptable system. In this thesis a computer model of such a system is put forward. The next chapter describes the signal processing tools and the system overview. Consequent chapters establish the working and validity of this model. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 Chapter 2 System Design and Methodologies 2.1 Objective The intended research work has been focused on modeling a novel sound delivery system that comprises of a microphone (FG 3329), a digital hearing sound processor (TI DHP 100) and an output receiver (Cl 8409). The novelty of the system lies in the fact that user’s ears are not occluded so that the ear canal is at least partially open for directly receiving ambient sounds. The microphone picks up the sound at a user's ear, amplifies, filters, and processes it. A transducer directly delivers this processed sound into an open ear canal. The transducer could rest on the ear pinna like a BTE module or may be implanted inside keeping the ear canal open. This amplified sound energy is then mixed with sound delivered naturally and directly into the ear canal. This technology hopes to reduce the acoustic feedback that can be described as a whistling, screeching, or ringing. Velcro Head Band for “ attaching microphone FG 3329 Microphone C l 8409 Receiver Plastic tubing attached to receiver sound port, inserted into ear canal Figure 9: Prototype Sound Delivery System Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 This acoustic feedback is often caused due to either acoustic leakage or the mechanical coupling between hearing aid receiver and the microphone. However in case of an open ear acoustics there are many techniques available to cancel out or at least perform band- limited cancellation. One such technique is adaptive feedback cancellation, developed by House Ear Institute. It requires the use of fast acting components that allow for rapid identification and elimination of feedback. Such a device has a large range of other advantages that include cosmetic appeal, reduction in discomfort level, reduced pain in ears, reduced ear wax, no rashes in ears, no dizziness etc. It is expected that such a system should also have a fast processing time (<10ms). This is important because of the mixing taking place between the amplified, delayed sound with the natural sound in the open ear canal. A loss of signal occurs at low frequency because some amount of signal would leak out through the open ear. Also if the transducer is a hermetic diaphragm type the output will be certainly limited at the low frequency where displacement for significant sound pressures levels can get large. We can expect that signals below 30 dB may not be properly amplified or may be completely lost in noise. The efficient and systematic development of such an implanted auditory prosthesis necessitates the use of computer models for understanding the prosthesis itself and its behavior in the intended application. The computer models play an important role in understanding electric, acoustic and electro-acoustic properties of the system under test. These models, simulations, and databases allow in experimenting different ideas about acoustics of perceived sounds and speeches in ways difficult to work if we had to depend on using actual physical systems. In this thesis, a time-domain non-linear auditory model Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 is developed for assessment of an open ear canal sound delivery system. The understanding of the system depends significantly on the effects of the external ear, which include head (skull), torso, pinnae, and ear canal on the Head Related Transfer Functions (HRTF’s). Essentially these measurements summarize the direction-dependent acoustic filtering a free field sound undergoes due to the head, torso, and pinna. These transfer functions aid in calculating the sound pressure that an arbitrary source x (t) produces at the eardrum. These acoustics effects have been modeled using KEMAR (Knowles Acoustics Manikin for Acoustic Research). KEMAR is used to make sound measurements that simulate those that would exist at the eardrums of a normal listener. The electronic components in the system add time delays due to processing; non-linear effects and other significant changes in perceived output due to varying gain controls. These become important to assess when we analyze the effects of mixing varying time delays in the amplified sounds with natural sounds. In physical systems, delays variable from 0 to 10 ms are acceptable to the user of an auditory prosthesis. The output transducer, microphone noise floor and acoustic feedback pose other limitations on the dynamic range of system. The auditory model described in this thesis is designed to match the processing within the human auditory system as closely as possible. Also to simulate the dynamic behavior of the system completely the room and speaker acoustics effects have been measured separately and incorporated into the model. The model has been validated using hearing aid fitting rules to theoretically optimize the outputs for individuals with varying hearing losses as measured by their audiograms. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 This sound delivery system utilizes direct acoustic stimulation of the ear and delivers the amplified sound at the eardrum. This model is able to predict the acoustic signal that will arrive at the subject’s eardrum, given the source signal and direction of arrival. The model includes two paths: unprocessed (direct from source to eardrum) and processed (via hearing aid). There are certain inherent delays that may be associated with the sound delivery system, which consists of a few analog parts and a few digital parts. The transfer functions of both these paths vary and thus alter the sound output. A simplified block diagram of the system is presented in figure below. Input White Room Noise Effect Acoustic X ’fer Func Source to MIC Noise generator Acoustic X fer Func Source to TM ► TM Acoustic Xfer Func Rcvr to TM TI Processor Saturation Analog Amp Analog Amp MIC, Receiver Feedback Path Figure 10: System overview The system once ready to implement would be capable of producing adequate sound pressure levels in the ear canal for patients with moderate and sever hearing loss, keeping their ear canal open. However the sound delivered by the system would be radiated out of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 the ear canal and eventually reach microphone causing feedback which has to be cancelled appropriately. Our proposed model of a novel open ear canal sound delivery system consists of the following elements: Acoustic Elements Direct source-to-eardrum path (direction-dependent) - This transfer function represents the acoustic properties of the ear canal. It is marked as 1 in the figure above. The acoustic input signal to the ear is function of numerous variables. It could be affected from head size, head shape, pinna, torso and clothing absorption. The ear canal tube acts as a resonator and typically resonates at ~2.5Khz. This is evident in the ear canal responses presented in the next chapter. This response is highly direction dependent. Source-to-microphone path (direction-dependent)-This path is a measure of the path from the sound source to the microphone of the hearing aid. The sound pressure level at the microphone is affected by the torso because of the constructive and destructive interference between the wave scattered by the torso and the direct wave to the ear. This response is highly direction dependent. It is marked as 2 in the figure above. Receiver-to-eardrum path-This transfer function is the measure of the acoustic path from the hearing aid receiver to the tympanic membrane. The acoustic properties of this path in the open ear canal are quiet different from that of ear canal occluded with hearing aid or an ear mould. The transducer in this study is mounted at tip of stiff, pliable wire and plastic tubing attached to receiver sound port. This would be inserted into subject’s ear canal and the tubing is slightly curved to follow bend of ear canal. It is marked as 3 in the figure above. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 Receiver-to-microphone feedback path - Feedback results from the acoustic coupling between the hearing aid loudspeaker and microphone especially at high gains. The main causes of acoustic feedback are high signal leakage (e.g. caused by a badly fitted ear mould or laughing) and a large acoustic feedback path (e.g. caused by bringing an object close to the ear). Acoustic feedback suppression in hearing aids becomes important since it reduces distortions and increases the maximum gain of the hearing aid. These feedback signals are eliminated by an adaptive digital feedback that generates an anti-phase feedback signal to partially or completely cancel the feedback signal. Acoustic feedback in this case is time varying. It is marked as 4 in the figure above. Electric Elements Texas Instrument processor-The compression/expansion and feedback cancellation processing is implemented on the Texas Instrument DHP 100 digital hearing processor that contains the TMS32OV5402 DSP chip and the Code Composer Studio ™ (CCS) software development environment which interfaces this battery powered stand alone card to the PC. Physically it mainly consists of DSP-CPU motherboard and a coder- decoder daughter card. The processor can be adapted to perform DGC, spectral shaping, noise filtering etc. The figure on the next page shows the block diagram of DHP 100 EVM. The processor has selectable sampling rates although we have use 16 KHz in this project. It uses 16 bit processing. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 \j- * ~ -y JX A G Address* JTAG HPI Figure 11: DHP 100 EVM Block Diagram Overview Electro-acoustic (Transducers) Elements Microphone-In a hearing instrument the most important components are the electro acoustic elements, microphone being one of them. Microphone picks up the acoustic signal and converts it to an electric signal for further processing. Different kinds of microphones are used in hearing instruments such as omni-directional microphones, which is essentially a pressure sensor in that the diaphragm is moved exclusively by sound pressure. A directional microphone has two sound ports and is capable of reacting better to sound from particular direction. In this research work an omni- directional microphone FG 3329 has been employed. The frequency response is shown in the appendix. Receiver-The function of a hearing receiver is to convert the amplified electrical signal back to acoustic signal. This is the final element in the processing chain. In this research work Cl 8409 is used. The frequency response is shown in the appendix. In order to determine the maximum amplified signals that can be delivered by the system it was necessary to identify receiver saturation levels. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 Frequency, Hz Saturation Level, dB SPL 500 95 1000 116 2000 >137 4000 >132 Table 1: Receiver saturation evels 2.3 Measurement and Analysis Methods All elements in the system have been modeled as Finite Impulse Response (FIR) filters. This is because coefficients of a FIR filter can be designed to closely approximate the desired frequency response specifications. FIR’s have a finite impulse response and are often referred to as non-recursive filters, convolution filters, or moving-average (MA) filters because you can express the output of a FIR filter as a finite convolution. In essence an FIR filter is a simple linear combination of a finite number of samples of the input signal. In general, an FIR system is described by the difference equation: M - 1 y(n) = ^ b k * x(n - k) k = 0 Or equivalently, by the system function M -1 H(z) = Y Jbk*z~k ( 1.0) ( 1.1) k - 0 An FIR filter is represented by its impulse response. The transfer function of the FIR filter is the Fourier transform of its impulse response. The filter output is the filter input Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 convolved with the impulse response. This technique is helpful in approximating the magnitude and phase response for all system elements. Thus, an FIR represents each system element. The filter coefficients of which actually represent the impulse response of that system (and in essence the frequency response). 2.31. Model Design using Weiner filters Wiener filters are a class of optimum linear filters, which involve linear estimation of a desired signal sequence from another related sequence. Given sample input and output signals for the system, the Wiener filter is the FIR filter that, when applied to the input signal, yields the best match to the output signal. Precisely, the Wiener filter is a time domain filter that designs the optimal FIR filter that approximates the system under test. The best-fit solution is taken to be when the sums of the squares of the errors are a minimum, where the error is the difference between a measured value and the calculated value (least-squares method). The software used to calculate Wiener filter is a C++ code with a graphical user interface that allows user to define two *.wav files: input and output and then computes a resulting *.flt file. This file is an ASCII file containing a list of numbers. These numbers are the coefficients of a filter whose z -transform is: H (z ) = f > z - ' (1.2) M -1 j = 0 Here N is the number of coefficients, bo, bi, bz... b n - i are the numerator coefficients. M is the number of denominator coefficients and ao, a\, & 2-. • a m - i are the denominator Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 coefficients. The software used to calculate wiener filter is a C++ code that has a graphical user interface as described in appendix. 2.3.2 Estimating Impulse Response of any System Weiner filter described above is used to estimate the impulse response of the system under test. It could be the acoustic response from the source till the tympanic membrane or response of any device including microphone, TI DHP 100. A hearing test device, which connects to the host PC via a USB, is used to play white noise to the system from its one channel and the other channel of the same device captures the response. This is described in detail in the next section on instrumentation and laboratory set up. These two *.wav files are then inputted in the Wiener filter program to compute an estimate of the impulse response of the system. A Matlab function is used to read back and plot this response. Hay w h i t e n a i mt from PC R e c o rd “ mpur* P C Figure 12: Estimating Impulse Response of any System Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 2.3.3 Deconvolvution Technique Some system elements cannot be measured in isolation. For example measured source-to- eardrum path also includes acoustic effects of loudspeaker and room. Undesired effects have to be measured separately, and then divided out (deconvolved). Mathematically, division of transfer functions is equivalent to deconvolution of their respective impulse responses. We developed a general-purpose technique to divide out any two-transfer function. Wiener filter design techniques were adapted to deconvolve one impulse response by another. This technique works by using the two impulse responses (input and output) to filter a noise signal. The two filtered noise signals are then again submitted to the Weiner program that designs an impulse response, which approximates the deconvolution of the numerator by the denominator. The above steps are computationally intensive and are performed using Matlab Signal Processing Tool box. Figure 13: Deconvolution Technique Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 2.3.4 Estimation of Feedback The adaptive digital filter (ADF) technique can eliminate the instability of the hearing aid by generating any anti-phase feedback signal from hearing aid output to fully or partially cancel the feedback signal. This acoustic feedback is time varying and the feedback canceller works by estimating the feedback path. The response coefficients generated by this filter are used to represent coefficients of an adaptive FIR filter in the simulated model (and thus the transfer function of the feedback path). This FIR filter is a 32-tap filter. 2.4 Signal Processing The incoming signal goes through the following signal processing: 1. The incoming signal is split into two parts and goes through two separate paths: processed and unprocessed. This is done after mixing the signal with room effects. 2. The unprocessed path augments the incoming signal in the same way as the physical ear (~1 to 1.5 dB gain) without any supplemental amplification except that provided by the ear canal. This would assist the user in way so that the ear canal is at least partially open for directly receiving ambient sounds. This part is modeled in the Simulink model as FIR filter whose coefficients are the impulse response of the measured signal. Finite Impulse Response (FIR) filter approximates the measured frequency response of the acoustic path from sound source to the tympanic membrane. The response is presented in next chapter. 3. The other part of the signal goes through complex parts of the sound delivery system starting with an omni directional microphone FG 3329. This path Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 represents the acoustics from the sound source till the microphone and is approximated by implementing FIR filter of 150 taps. The FG 3329 has a sensitivity of about 1.0V/0.1 Pa relative to -53 dB SPL, which translates to a fixed gain of 0.0224. 4. A fixed noise floor of 30dB SPL is incorporated into the model. This was chosen to be low enough to ensure that no clipping or saturation occurred anywhere in the system, but high enough to ensure that signal is well above the ambient noise floor of the sound room. This represents the microphone noise floor. 5. A saturation block implements the microphone FG 3329 saturation, which limits the maximum signal in the system to be 103 dB SPL. This is to guarantee that DHP 100 would not generate any distortion at its output, which would then drive the receiver incorrectly. Per specifications and laboratory verification the D/A converter of the DHP 100 would begin to saturate at 103 dB. 6. House Ear Institute’s processing algorithm is applied via the DSP sound processor and its programmable chip. The system has been configured to provide adequate expansion-compression and band limiting to the incoming signals. It is capable of sampling at a rate of up to 32 KHz and the platform can be programmed to apply various noise reduction and spectral shaping techniques. Some of them have been implemented in this model and described in the model validation chapter. Cl 8409 is a high sensitivity receiver with a wide bandwidth response. It converts the electric signal coming out of the TI DHP100 into acoustic signal. Cl 8409 has a sensitivity of about 93.2-mV/Pa. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 Chapter 3 Research Design and Methods 3.1 Scope of Thesis This thesis describes the simulation design methodologies and results for a novel open ear canal sound delivery system. Since the ear canal is open there are two signals that arrive at the tympanic membrane: the unprocessed (direct from source to eardrum) and processed (via hearing aid) signals. The research work investigated the effects both physical and electric parameters on the final output of the sound delivery system. Some of the effects studied in this work include those of: 1. System gain: This gain represents the overall gain of the system, including that of the ear canal and all the electronic components like, microphone, TI DHP 100 sound processor, receiver etc. This gain is variable and depends upon the settings of the TI DHP 100 which can be programmed to provide a higher gain in case of severe hearing loss. Also depending on the microphone specifications this gain may vary by a few dB. 2. Processing delay: This delay arises due to the novel method of sound being delivered at the tympanic membrane. The signal arriving at the tympanic membrane is a sum of both the processed and the unprocessed path. The delays due to the electronic components and those due to head shadow effects have a substantial effect on the final output. 3. Open ear canal losses: Since the ear canal is open all the time there are certainly some losses at low frequencies. These could be equated to the signals of the levels of the noise floor of microphone. They were important to study to understand the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 amplification and distortion present when the input to the system is as low as noise floor. 4. Transducer limitations: The transducer is one of the most important components of such a sound delivery system. It essentially delivers the processed sound at the tympanic membrane. Its effects are important to study because this model forms the basis of a procedure that could be used to characterize any new transducer. It is a scalable and modular system making investigation of any new components fairly straight forward. This model developed would be able to predict the acoustic signal that will arrive at the subject’s eardrum, given the source signal and direction of arrival. It has been identified that the resulting acoustic output is limited in performance by the maximum output the transducer can deliver as a function of frequency. Feedback has been reduced as much as possible by the use of high quality feedback reduction software. The microphone noise floor and saturation limit the dynamic range of this system, which is roughly around 75 dB. Processing delays were varied within the electronic board and its affects understood. All electro-acoustic measurements along with the feedback canceller algorithm running were made by fitting the prototype in the ear canal (KEMAR). A measure was identified for deciding maximum stable gain the system is able to provide. The signal processor used included the forward path hearing aid amplification and the feedback canceller algorithm (that was turned on or off using host PC). The research work has been performed only for monoaural hearing loss but is definitely capable of predicting results for a binaural hearing loss condition. This remains a future exploration. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 3.2 Instrumentation The measurements of all the transfer functions have been carried out on Knowles Electronics Manikin for Acoustic Research (KEMAR). It is manikin for hearing aid acoustic research with median human adult dimension. Ear simulation matches the acoustic response with an auricle, an ear canal, and eardrums that equal the median ear in dimension, acoustic impedance and modes. Its dimensions of torso and head are based on the published anthromorphic data. Validating results have been published in journals and technical reports presented by manufactures. This appropriately proportioned and designed manikin provided researchers with lifelike test conditions and experimental flexibility. The other necessary instrumentation used in the study is: Hardware: This consists of the following components: Hearing Test Device (HTD): It is a USB based playback device with a built in programmable amplifier. It is capable of generating various high quality test signals at a wide range of presentation levels, generally used for air/bone conduction testing. HTD (generally inside of a test room) is controlled via a USB interface by a host PC which is usually outside of a test room. It has two input channels (left and right) and is capable of performing both mono and stereo recordings. For the output part, it supports three types of transducers: two speakers, two earphones and one bone conductor. It has been used for playing white noise through various test systems and capturing the capturing their impulse responses. Preamplifiers: Preamplifiers permit efficient signal transmission over ordinary coaxial cables. A preamplifier developed by Etymotic Research Inc. is used in this work. It provides the highly stable DC polarization voltage required by precision microphones Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 along with low noise amplification enabling it to drive input and output long coaxial cables. It is rugged, lightweight and it runs on a 9V battery inside the case. Battery power and metal box keep external noise and interference sources away. Sound Level Meters (and measuring amplifier): B&K Type 2609 is used in frequency range with principle application as a microphone amplifier. The meter measures true RMS voltage and also includes A-weighting network. In this research work it has been used primarily as an input amplifier when recording sound via speakers. It could be used as an audio frequency compressor amplifier, in multi channel noise monitoring systems. It is capable of providing a gain of as much as 90dB. •4--------------------------------------Meter Panel Gain setting controls Figure 14: B&K Type 2609 Sound Level Meter Reference Microphone: This is a Larsen Davis Model 2575 random incidence 1-inch microphone. It exhibits good temperature stability, an important parameter for microphones. The Model 2575 is the appropriate microphone for use with precision sound level meters intended. A random incidence microphone is designed essentially to measure sound pressure levels in a diffuse sound field, as they existed before the microphone was introduced into the diffuse sound field. Its response essentially represents the combined effects of the room, speakers, floor and walls. They are presented in the subsequent sections. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 Microphone FG 3329: The microphone is a Knowles FG-3329 microphone, which is enclosed in BTE case. The microphone is mounted directly above manikin’s pinna and is held in place with Velcro headband. .:(< ' I B ’ T <- Mic FG 3329 -----------------Connector to TI DHP Figure 15: Microphone FG 3329 TI DHP 100: The sound processor has an analog input section (microphone preamplifier) combined with an A/D converter. It houses a TI TMS320VC5402 fixed-point DSP chip along with a D/A converter used to convert back the signal. It consists of an analog output section (receiver driver), which drives the output receiver. It has selectable sampling rates however in this study 16 kHz has been used for all measurement purposes. Output of T I' DHP 100 (To Transducer) Input to TI DHP 100 (From Microphone) Power Supply Processing Board Unit Figure 16: TI DHP 100 Board Output Transducer: It is a Knowles CI-8409 receiver. It is mounted at tip of stiff, pliable wire with plastic tubing attached to receiver sound port, inserted into ear canal. The tubing slightly curved to follow bend of ear canal. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 Connector to TI DHP 100 Cl 8409 Receiver Bent Plastic Tubing H i Figure 17: Knowles CI-8409 Receiver Hardware Performance Levels: The potentiometers in DHP 100 analog input section control maximum input level before A/D converter clipping. These have been adjusted to clip at 103 dB SPL input to mic (for 1-kHz tone). Also to avoid clipping in internal DSP path, input level plus digital gain cannot exceed 103 dB SPL. Signal processing software: This software drives the TI DHP 100 sound processor and interfaces its TMS32OVC5402 battery powered card to the PC. The processor has three bands of compression/expansion. These are at defined at three bands of frequency ranges of 500 Hz, 1000 Hz and 4000 Hz corresponding to the low-, mid-, and high-frequency bands, respectively. Compression is a functionality of automatic gain controller where the gain reduces as the signal level increases. A multi-band compression algorithm compresses the sound such that audio signal is no longer distorted in the auditory system and the perception is restored to a louder level. The compression parameters that can be specified are ratio, threshold (in dB relative to maximum level), and gain/attenuation. In the model the Dynamic gain Compression (DGC) block evaluates two independent gain functions. The resulting of the two gain functions, plus an additional fixed amount of gain (or attenuation), are all summed to form total gain/attenuation. The figure below depicts Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 block diagram for the software. Compression applied between noise floor and 95 dB SPL where as expansion applied below noise floor. IN + Feedback Canceller 4 Lowpass (<1 kHz) -> Compress/ Expand Bandpass (1-3 kHz) - ► Compress/ Expand Highpass (>3 kHz) -* Compress/ Expand Atten Atten — ► Atten Limit Pitch — ► Shifter- -> OUT Figure 18: Block Diagram of Signal Processing Software The software is also adapted to include a feedback canceller and a pitch shifter. The feedback path is simulated with a FIR filter which is derived from a laboratory measurement. The feedback canceller can be turned on or off depending on the threshold of hearing, such as in case of high gain instability sets in even with feedback canceller turned on. The purpose of pitch shifter is that it reduces artifacts caused by feedback canceller when listening to music. However the software tool described is limited by only 8 choices of compression ratio and gains and attenuations quantized to 1.5-dB steps. 3.3 Azimuthal Dependence of Head Related Transfer Functions (HRTF) The direction-dependent frequency response of the ear is called Head-Related Transfer Functions (HRTF’s). An HRTF is a measure of the acoustic transfer function between a point in space and the eardrum of the listener. The HRTF includes the high frequency shadowing due to the presence of the head and torso, as well as directional-dependent Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 spectral variations imparted by the diffraction of sound waves around the pinna. HRTF’s summarize the direction-dependent acoustic filtering a free field sound undergoes due to the head, torso, and pinna. Typically, HRTF’s are measured from humans or mannequins for both the left and right ears at a fixed radius from the listener’s head. HRTF’s vary with different azimuths left-right direction and front -back direction, which can be measured in degrees or radians. In our research study the zero degree sound sources is the direction is designated as being directly in front of the manikin. 90 0 correspond to the location of the active ear (from where ear drum signal is taken). Thus a 90 0 orientation means that active ear is closest to sound source. At 270 °or -90 °(also called contralateral) the microphone to the source is closest and active ear is on far side of head. These are specified as minimum phase FIR filters. Azimuthally directivity was maintained by rotating the manikin around the vertical axis while keeping the acoustic source stationary. ■ Source Sound in ten sity at rfefciear -90 0 -+ 90 D irectly opposite left e a r D irectly in fi-oni o f subject D irecfly opposite rif lr t e a r Figure 19: Azimuthal Dependence of HRTF Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 3.4 Calibrations The model needs to be appropriately calibrated to make adjustments for any differences in measuring methods; different sets of equipments used and also change in azimuths. It Voltage Voltage TI DHP 100 DSP A/D D/A Rcvr is also required to calibrate all the electric and acoustic signals coming in and out of the various instruments in the physical system and to set up reference levels for various measured quantities. Figure 20: Calibrations The system elements require the calibrations convert the dB SPL units into appropriate voltage units and then these units back into dB. Microphone Sensitivity: The sensitivity of a microphone is defined as its output level for a reference sound pressure level. The most common reference sound pressure level is 1 Pascal or 94 dB SPL. A pure tone of 94 dB SPL is played at the Microphone and the voltage coming out of the Mic is measured. Voltage = 6.4mV The sensitivity of FG 3329 per datasheet is 53 dB relative to l.OV/O.IPa Since lPa=94dB .‘.0.1 Pa=74 dB spl, Sensitivity of Mic -1.0V/ (74+53) = 1.0/127 dB spl Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 Electrical Gain of TI DHP 100 Sound Processor: Voltage I/p = 6.4 mV Voltage o/p=359.5 mV Electric Gain =201ogl0 (359.5/6.4) = 51 dB O/p Transducer Sensitivity: A full-scale deflection at the TI is equivalent to 2.3 mV, Corresponding output voltage =234 mV dB spl reading on Measuring Amplifier =102 dB spl lPa=94dB spl ?= 102 dBspl = (10)8/2°pa = 2.51 Pa But 102 dB spl=234mV Also 2.5 lPa=234mV Thus Output Transducer Sensitivity=93.2 mv/Pa 3.5 Acoustic Transfer Function Estimation The transfer function provides an algebraic representation of a linear, time-invariant filter in the frequency domain. To compute transfer function we use deconvolution approach because Wiener filter approximates the impulse response of the system, which are otherwise impossible to measure in isolation. The signals are acquired a 2 channel recordings, with HTD loop back in one channel (Input wav file) and microphone signal (Output wav file) in other channel. Weiner filter also models the absolute delay associated with the propagation of acoustic signals in different azimuths. The distance Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 between center of KEMAR head and speaker is lm (~3 ft) and sound travels across at 330m/sec. An approximate delay of 3ms (72 samples) is included in all impulse responses. A further delay is added or subtracted depending on the azimuth. For example delay for ipsilateral is shortest at 65 samples and contralateral (away) is maximum at 87 samples. FIR filters designed ranged from 150 taps for KEMAR HRTF’s to 200 for FG Mic response. A filter of 100 taps was set up for TI DHP 100 response measurements I F S I U I E R A L I>*1»V IJftU S BACK Delay: Jifin is L \ D ista n c e * * lrn FRONT Delay: 3.1£uts CONTRALATERAL Delay: 3.61ms Figure 21: Delays in the four azimuths A series of measurements was made to characterize and optimally configure the system. To make measurements, KEMAR was placed in the center of a sound booth, facing the Tannoy speaker, which delivered test signals (white noise). The FG 3329 and the sound delivery system were mounted on KEMAR’s left ear. The two microphones used were V 2” KEMAR ear canal microphone and FG 3329 mounted on the pinna. Each microphone was connected to a sound level meter that was calibrated to allow direct reading of sound waves. The first microphone measures approximate signals at a typical ear drum while Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 the second microphone is intended to determine signal at the input to the sound delivery system. Head Related Transfer Functions (HRTF) As already described before, HRTF’s are filters, which describe the acoustic filtering of the head, torso, and external ear, or pinna, perform on a sound. For this research work we focused only on four sets of azimuth that have the most effect on the acoustical characteristics of the open ear canal sound delivery system. The lab set up is fairly simple with KEMAR microphone connected to a pre amplifier and a then to a sound level meter to read the sound level. Both input and output wave files are supplied to the Weiner filter program to compute responses. Pre- Amplifier Sound Level Meter HTD Figure 22: Laboratory set up for ear canal HRTF measurements jjjjjl/y y v v w v — 0 50 100 150 CONTRA 50 100 150 50 100 150 Figure 23: Impulse Kesponses for ear canal in all 4 directions Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 The speaker was kept in the same position and the manikin was rotated in four directions. The above figure shows the impulse response of the ear canal at all four azimuths. It is observed that the response is higher for ipsilateral HRTF’s than that for contralateral HRTF’s. In this case, a lower response is possible, since the contralateral ear typically receives less power than the ipsilateral ear during the measurement process because of the head shadow effect. HRTF data can be understood in the frequency domain very well, where the magnitude response of various HRTF’s is plotted as a function of frequency. In the following figures a Fourier transform of the impulse response of the acquired signals. Matlab command ‘FREQZ’ was used to plot the transfer functions. Command [h,w]=freqz(b,a,n,Fs) returns the frequency response vector h and the corresponding angular frequency vector w for the digital filter whose transfer function is determined by the (real or complex) numerator and denominator polynomials represented in the vectors b and a, respectively. Fs is the sampling rate while a is 1 for a FIR filter. The response is calculated at n angular frequencies between 0 and 211. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 2000 3000 4606 5000 Frequency (in Hz) Frequency (in Hz) P h ase deg rees -15000 •15000 2000 3000 Frequency (in Hz) 6000 7009 3000 IB S 5000 Frequency (in Hz) 6000 6000 2000 7000 Figure 24: Front and Back Transfer Functions Since it is proposed that pinna filtering provides spectral cues for sound localization, there is limited front-back distinctions in the HRTF’s (or more aptly in this case “Free- Field-to-Eardrum Transfer Function” or FETF).In general these differences could be uniquely determined by time differences along with the pinna of the outer ear reflecting sound differently for the positions around the listener. As evident from the response it is difficult to distinguish between sounds in front of the head and the "mirror image" position behind the head (i.e. +0° and +180°) without additional information.The units of the magnitude response are Power Spectra Density (PSD). Matlab function psdplot(Pxx,w) plots the power spectral density data contained in the vector Pxx at the frequencies specified in the vector w. The vectors Pxx and w must have the same length. The magnitude of the data vector Pxx is plotted in dB per Hz per sample versus frequency w on one plot. The units for frequency on the plots are Hz. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 Magnitude In dB a ■ « 0 m m 3 0 Q Q 4 Q 0 @ 66Q0 6000 Frequency in Hz M l 7000 8000 P h ase In d e g reH ■15000 2000 5000 8000 m m m Frequency in Hz Frequency in Hz Figure 25: Contralateral and Ipsilateral Transfer Functions For some frequencies and incident angles, the head has an amplifying effect on an incident plane wave at certain points near the head due to diffraction. There are some locations on the contralateral side of the head where this effect occurs, even though the head directly blocks or shadows the contralateral ear. These are often referred to as diffraction effects and they can be seen most easily in shapes in the contralateral HRTF’s corresponding. The head has a slight amplification effect on lower frequencies due to diffraction on the contralateral side. In addition, amplification effects are expected and can be seen for high frequencies in ipsilateral HRTF’s due to reflection and the ear’s proximity to a rigid surface (e.g. the head). Elevation effects in the ipsilateral HRTF to some extent may be due to the pinna. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 Source- to- microphone response (FG3329 omni direction microphone) These impulse responses represent the acoustic properties of the source to microphone path. The microphone is mounted on KEMAR’s head wearing a Velcro headband, directly the subject’s pinna, on the same side as the sound delivery device (receiver) with the look direction pointing forward. The microphone is connected to the processor, output of which is connected to the Cl 8409 receiver. A white noise is played and corresponding output is captured through HTD and fed to Weiner filter program. TI DHP 100 > C l 8409 Connection from FG MIC ► Sound Level Meter-2 ^ Pre-Ampufier-2 p Pre-Amplifier -1 p Sound Level Meter-1 Figure 26: Laboratory Set up for FG 3329 Response Measurements FG M IC R«»p fflONT 20 40 FG N 60 80 100 120 140 1 C Resp C O N TR A IP SI Figure 27: Source-to-microphone response for 4 directions Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 X Axis: Number of Samples; Y Axis: Absolute Magnitude of Response TI DHP 100 Processor This component is a purely electric one however it still needs to be characterized in terms of its impulse response. The response of the processor is measured as a standalone system. This processor also drives the transducer with its sound delivery tube pointing in straight in the ear canal. A separate power supply and HTD adapter is used to supply power to TI processor. A power supply board supplies power to HTD Adapter which splits into appropriate positive and negative lines. The output of this adapter forms the input to the TI processor. This is necessary to avoid any over driving of the card with high voltages. The output connects to a measuring amplifier (often referred to as a sound level meter) and the sound pressure level is read off the meter. The input is a 65 dB white noise which along with captured output wave file is supplied to Wiener filter to compute a response best matching the response of TI processor. The processor drives the output transducer and the TI response is divided out to get exact output transducer response, which is shown next. All measurements obtained using white n Sound Level Mcter-1 H TD HTD ADAPTER Gaussian noise. Figure 28: Laboratory set up for measuring response of TI DHP 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. N um ber of sam ples F requency in Hz Figure 29: TI DHP 100 Response O/P Transducer Impulse Response The prototype sound delivery device is a Knowles Cl 8409 receiver mounted at the end of a stiff pliable wire. Plastic tubing attached to the receiver and the device is positioned so that tubing is partially inserted into the subject’s ear canal. The Texas Instrument DHP 100 output connects to the sound delivery system and thus drives it. The output transducer is captured along with the TI processor response and its response is divided out using the deconvolution technique. The transducer response matches very closely as obtained from the data sheet. The data sheet is presented in the appendix. Magnitude in dB P h ase in degrees Number of sam ples Figure 30: Cl 8409 Receiver Response Frequency on Hz Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 Reference Microphone Response The reference microphone response essentially contains the room and speaker effects. The HRTF’s are recorded in an anechoic chamber; however there is still a possibility of some room response being present in the acoustic transfer functions. These could be due to reflection off the floor, windows and walls of the room. The best way to measure this was by removing KEMAR from the room and placing a reference microphone (in this case a Larsen Davis Model 2575 random incidence 1-inch microphone) in the center of room. A white noise input and its corresponding output and supplied to the Wiener filter create the response which is essentially the room effects. These have been incorporated in the model to simulate the physical system and hence validate the model as closely as possible. Num ber of sam ples Magnitude ifl dB . . . . . . . . . . . . . . . . . . I ' " ' ! I . J . . . ^ / j | j 1000 3 0 0 3000 4000 5B Q Q Frequency in Hz t o o o s e o o P h ase in degrees 3000 4000 Frequency in Hz Figure 31: Reference Microphone Response Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 3.6 Model Summation and Results As the sound travels thru the ear canal it undergoes slight amplification and the ear canal resonance is evident at about 2-2.5Khz. The other half of the system at the same time is filtering and amplifying the same sound and delivering it at the tympanic membrane. The input to the model was white noise at 65-db spl. The physical measurements were taken with the KEMAR and similar white noise was given as input to the computer model. The physical system set up is shown below in the block diagram. The results from the physical system were then used to compare y=the results of the Matlab model. In almost all the cases the results closely match with the simulation results. Upper Path TI DHP 100 C l 8409 Connection from FG M IC ► Pre-Amplifier-2 ► Sound Level M eter-2 Pre-Amplifier -1 Sound Level Meter-1 Lower HTD 1 1 1 1 4------------------- — ► Figure 32: Laboratory set up for measuring the combined response Front Azimuth Summation Results The response of the system is similar to the ear canal for the upper path of the system. The lower arm with the complex sound delivery system provides a 30dB gain to the incoming sound. The graphs below explicitly show the amplification and changes the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 filtered sound goes through. The other results shown are the actual lab measurements and the simulated results through the system built in Matlab. U p p e r p a rt o f th e m o d e l 2 0 C 150 100 50 0 , F r e q u e n c y L o w er p a rt o f th e m o d e l C D 150 1 1 0 0 2 1 Q_ 2000 6000 8000 F r e q u e n c y C O M B IN E D F R O N T R E S P O N S E S IM U L A T E D 200 m 2 , 'e 150 S E 100 t 5 C L F r e q u e n c y R e s u lts a s o b ta in e d in th e I: 200 S ' | 150 f 100 2 Q. 6000 8000 2000 F r e q u e n c y Figure 33: Front Azimuth Summation Results Back Azimuth Summation Results The instrumentation for these remains the sane as for Front azimuth however the manikin is rotated to be at the back of the speaker. The responses should be almost the same due to front-back ambiguities. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 U pper part of th e m odel e g 2 6 50 i IX 6 000 F req u en cy Lower part of th e m odel C D I o . 6 0 0 0 U U M tJIN tU fcdAUK H b S P U N S fc S iM U L A Ib U C D 50 i 2 0 0 0 4 000 F requency R esu lts a s obtained in th e lab C D 2 E 2 • Q 50 Q. F requency Figure 34: Back Azimuth Summation Results Ipsilateral Azimuth Summation Results When a sound reaches the ears, the ipsilateral ear, which is closest to the sound, perceives it first and the sound is louder than at the contralateral ear, which is further from the sound source. There are little sharp peaks and notches in this response due to limited variability. There are low head shadow effects as compared to the contralateral ear. Upper pert of the model COMBINEO IPSI RESPO N SE SIMULATED 2000 4000 6000 B O D O Frequency Lower part of the model 2000 4000 6000 B O O O Frequency Results ae obtained in the lab 200 2 i 6 w 1 S 0 100 50 i IX 0 4000 Frequency 0 2000 2000 4000 Frequency Figure 35: Ipsilateral Azimuth Summation Results Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 Contralateral Azimuth Summation Results The manikin is rotated by 90° such that ear with microphone faces away from the speaker. As the sound source moves away its responses become much more complex for the contralateral ear than the ipsilateral ear. In this case a series of grooves and notches accentuate or suppress mid and high frequency components in an azimuth dependant manner. The high frequency shadowing is he reason for this complexity. U p p e r part of th e m odel C O M B IN E D C O N T R A R E S P O N S E :S IM U L A T E D 200 m T 3 cu -a 3 150 C CD C O 1 1 0 0 3 u ( D a . C O C D J o Q_ 2000 4000 F re q u e n c y 6000 0000 L o w er part of th e m odel 200 CD T 3 -S 150 3 C 0 5 0 ) g 100 3 C D O J § Q Q_ 2000 4000 F re q u e n c y 6000 8000 200 S ' T > -S 150 3 C 0 5 C D g 100 3 L > O ) a . C /3 C D i 2000 4000 F re q u e n c y R e s u lts as o b ta in e d in th e lab 200 CD -a -S 1 5 0 3 ,-ts c CD < a g 100 3 ■ G a) a. C /3 I o C L 4000 F re q u e n c y 6000 8000 Figure 36: Contralateral Azimuth Summation Results 3.6 System Limitation: Microphone Noise Floor and Dynamic Range The measurement of the Total Harmonic Distortion (THD) becomes essential because it indicates the magnitude of harmonics and other frequencies not present in the initial signal. These output frequencies not present at the input signal are essentially a measure Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 of the non-linear effects present in the system. The most commonly used test signal in such measurement is the single sine wave, which has been used in this work. It is well- understood fact that the property if a non-linear device is its ability to produce new frequencies. This energy is created at other parts of the spectrum. In this model there is a constant noise floor set at 30dB to simulate the microphone noise floor. To measure the Total Harmonic Distortion (THD) of the system a range of pure tones starting from as low as 20 dB SPL were given as the input to the model through 150 dB SPL. The pure tone frequency was maintained constant at 1 KHz. Essentially this is a measure of the THD + Noise since the noise floor is not subtracted of the calculations for THD percentage. The model shows an increasing strength of harmonics as the signal strength increases. However at lower input levels it is predominantly noise and as the signal strength increases beyond 103 dB (Saturation level) there is a loud increase in the signal energy. A graph of % THD Vs input dB SPL is presented below. 260 200 150 o X t- 100 50 °20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 Purs Tdhs Inout in DB anl(RMS) Figure 37: Percentage THD Total Harmonic Distortion a s a function of input do Spl Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 The figure below shows the system response to a pure tone of 103 dB spl and the response of the same system at 104 dB. The strength of these harmonics increases as the signal strength is gradually increased. The maximum THD is as high as 250%. Pdwer Spectrum I Magnitude in dB 1 -... V - m m - m & a m Frequency in Hz Frequency in Hz Figure 38: No Harmonics at 103 dB Figure 39: Harmonics at 104 dB Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 Chapter 4 Model Validation After accurate demonstration of summation of the processed and unprocessed path the next and concluding step for the computational model is establishing its validity. Validation of this modeling study is accomplished by evaluating goodness-of-fit between predicted and measured acoustical parameters. These parameters range from the different azmiuthal acoustic transfer function sets (ipsilataeral, contralateral, front). The digital sound processor is also capable of providing various sources of non- linear effects required to study the behavior of the model under such conditions. These include dynamic gain calculation, which represents the gain/attenuation calculation for multi band compression/expansion. A whole range of other range of fitting parameters is made use of in the validation study. These are derived from the hearing aid fitting rules modification of standardized fitting rules that provide theoretically optimal outputs for varying hearing loss conditions. Three hearing loss conditions were selected based on the audiograms obtained from a previous study conducted at House Ear Institute that evaluated the feasibility of the sound delivery system described in the thesis. It was established from that study that the sound delivery system was capable of providing adequate sound pressure levels in the ear canal of patients. The audiograms from the subjects were divided into three main categories of mild, moderate and severe hearing loss. This was done to span the whole range of hearing loss conditions and not just validate the model with predefined data. Three sets of hearing loss conditions for these in units of db HL were designed. This reference is based on a typical “normal” threshold at each pure tone frequency. dB HL reference changes with the frequency of the signal, but not with the receiver. (Also Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 important to note that 0 dB HL at 1000 Hz=6.5 dB SPL while 0 dB HL at 200 Hz=18 dB SPL) thresholds 500 Hz (dB HL) 1000Hz(dBHL) 2000 Hz(dB H L ) Mild 0 25 30 Moderate 23 41 51 Severe 36 65 63 UCL’s 68 86 98 Table 2: Reference Conditions for Validation The minimum sound pressure level at which a person can hear a sound at a given frequency is defined as the threshold of hearing. The table above shows that subject’s thresholds cannot exceed these limits, or required gain would cause instability, even with feedback canceller enabled. It also lists the uncomfortable level (UCL) that has to be kept in mind before applying gain. The normal thresholds are 0 dB at all frequencies. A Matlab script then calculates the compression ratio, target gain for each band, a fixed gain value and limiter gain. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 jl'Vhc*<W^4<\bS*4\dAuj3\dh8.out U«d«M(fftun J i* l2 i.S i OitptayAOF I rO W ««i»(Si«<W .............. 1-3kM iB.nl k ik H iB jrd i s * l < » ,« uiip ieo n ,5 .,w jaiijS i) | T ~ w < te | ( 1 w n » | p F " w « | l t » 8 j |5 W ritg | [2 w « n | | l M 3 W lile l Urafcefj^rridB.Mutocterf ? 5,rar»ge4gto»42{ f ~ « s d bm *« tecahtM (dB, tw » & i a t X tn & . -96 to 0) p 2 W ite j TgW I/O <feb« f t * rang* 1.3 !o 9.2) f l T * SM ^Q trr j Fewiiack-Canc***t'- (* U ft <" fight 0 « % (* a r> 9 ife t« 9 t2 tQ l* 3 | » W nte[ ; QC(**«r**geO p Write [ \ ( l « 0 3 ; 1 “ MeaaxementMode T EwbteFBC Pteh shift ampftucfe. s m ix (myd b« >« 0) P ith drift pood, seconds (mud b e ', 0! P 8*>d-*nted Cu«f(H*) |930 3 F “ , W « 9 R P t e t e d ^ B e t e e t e f - T lrothcM ltangaStal] j M V i » | tetf2rf«u(ranfl*-t8»431| Write [ DstalMt»te,i»|»9B255«l»>l j2 * 7 W i»»| ; Release rime, second (must b*>« 0) j* Write | I Forgetting facto * # t m (rs»ga 0 to 1 ^ p Write [ ; f lo fla p sip w e ic G range) 2 to 32j | T “““ Write | j f f ^ s ig W to detector |Hi^apawerroetegnteLetn) 3 | ? -R « fw s e to Pwiocfc Sgw te............... P EnabteptehjhRwiFFperiodc r DeaUelmdbacfc^wcetelFFpsiijdb l n e r t a s # Q C t v j lffp * » d fe Write Figure 40: GUI for Interface Software The feedback canceller can be turned on or off and so can be pitch shifter. However in this work, pitch shifter has been left on in all cases. The fitting parameters from the script are used to drive the TI DHP 100 software. The output of which further drives the receiver, which delivers the processed sound at the tympanic membrane. The lab setup essentially remains the same briefly described as: The microphone signals were pre-amplified by and then presented to the analog inputs of TI DHP-100 board. The board contains 16-bit stereo A/D and D/A (programmed to operate at 16 kHz) and a TMS32OV5402 DSP processor. The boards were programmed to perform the processing algorithms described above, and then apply frequency- dependent amplification appropriate for each subject's hearing loss. The analog output signal was presented to the subject via a Knowles receiver CI-8409, embedded in a BTE Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 module. It is apt at this point to talk briefly about the dynamic range of this system. The dynamic range is often mentioned as the difference in sound pressure level between the saturation or overload level and the background noise level of an acoustic or electro acoustic system (measured in decibels). For a sound or a signal, its dynamic range is the difference between the loudest and quietest portions. The human hearing system has a dynamic range of about 120 dB between the threshold of hearing and threshold of pain. Our system is capable of having a dynamic range of as much as 73 dB. This comes from the noise floor of microphone fixed at 30dB and the A/D clipping level fixed at 103dB (per data sheet). The House Ear Institute processing (compression/expansion and feedback cancellation) is implemented on the TI DHP 100 processor. The FG 3329 output connects to the DHP 100 processor, which in turn connects to the sound delivery device. Potentiometer on the DHP 100 control analog input and output levels. The processor supports stereo processing but in this research work only left channel of DHP 100 is used, regardless of which ear is tested. Thus the overall system gain as contributed by all the components is a variable gain. However the maximum gain the system provides is about 30dB SPL. The validation results are presented below. 4.2.3 Front Azimuth Validation Studies As sound passes from a free-field source to the ear canal, its spectrum is transformed by the interaction with the head and external ear. We saw in the previous chapter that this transfer function varies with the angle of incidence of sound, so the spectrum of the sound in the ear canal carries directional information. To perform validation studies we programmed the TI DHP 100 card to perform optimum gain to the hypothetical hearing Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 loss conditions. The data was analyzed in Matlab and a high correlation between physical systems is evident. The results are presented below. Y Axis represents Power Spectrum M ILD H E A R IN G L O S S M O D E R A T E H E A R IN G L O S S Lab Results — Sim ulated Lab Results — Sim ulated 13ower SpectreUft .Density in dB L -20 -20 -40 -40 2000 4000 6000 6000 2000 4000 Frequency in Hz 6000 boqo Frequency in Hz S E V E R E H E A R IN G L O S S — Lab Results — Sim ulated -20 -40 6000 2000 4000 Frequency in Hz B Q D O Figure 41: Front Azimuth with FBC ON Feedback canceller off By turning the feedback canceller turned off we are able to analyze how capable is the system in providing gain at condition in which there is increased constraint in the form of acoustic feedback. A benefit of feedback canceller is evident since the output in feedback canceller on condition is as high as 60dB while that in case of off is about 50 dB. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 F R O N T FB C O F F Lab Results Sim ulated Lab Results Sim ulated F ower Spectrum C ensity in dB 20 -20 -20 -40 -40 2000 Frequency in Hz 4000 6000 8000 2000 4000 6000 Frequency in Hz 8000 Mild Hearing Loss M oderate Hearing Lass Figure 42: Front Azimuth with FBC OFF 4.2.2 Ipsilateral Validation Studies The laboratory set up and hearing loss conditions remained the same. The manikin was rotated such that the ear under consideration was closest to the speaker. The feedback canceller was alternately turned on and off. Y Axis represents Power Spectrum in dB. MODERATE HEARING LOSS MILD HEARING LOSS 60 60 L8b Results Simulated Lab Results S i m u l a t e d 40 4D Power Spectrum Density in 20 -20 -20 -40 -40 6 Q 0 0 2000 Frequency in Hz 4000 6000 2000 Frequency in Hz 4000 8000 SEVERE HEARING LOSS 60 Lab Results Simulated 40 20 -20 -40 2000 4000 6000 Frequency in Hz Figure 43: Ipsilateral Azimuth with FBC ON Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 The results exhibit a high correlation in the low to mid frequency range with occasional notches and peaks in the laboratory results. These are not so evident in the mathematical model may be due to sharp adding and subtracting of in phase and out of phase signals. Feedback canceller off The feedback canceller when turned off presents an interesting result at frequency range above 6 KHz. This may be due to limited variability in head shadow effects at this high frequency. The system breaks into loud oscillations at high hearing loss condition and with feedback canceller turned off. This makes the measurement difficult and meaningless and hence omitted here. Y Axis represents Power Spectrum in dB. — Lab R e s u lts — S im u la te d 0 2000 4000 6000 8000 Frequency in Hz 60 — L ab R e s u lts — S im u la te d 40 Power 20 Spectrum ~ " in 0 Density in dB ■ 2 0 ■ 4 0 0 2000 6000 Frequency in Hz MILD HEA RIN G L O S S M O D H E A RIN G L O S S Figure 4: Ipsilateral Azimuth with FBC OFF It is important to note that the slight mismatches above 6 KHz are not very significant due to the roll off in human speech around this range and certainly due to transducer bandwidths which typically does not exceed 6 KHz. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 4.2.3 Contralateral Validation Studies Hearing with the era farther from the sound source as in case of contralateral is a subjective process. Testing any aspect of this requires special considerations since there are a wide range of parameters that affect its sound output. The data presented below is the comparison of the laboratory system and the computer model. A striking correlation occurs in this case with an exception of some noise in the laboratory data which is not surprising. The physical conditions have an important effect on the additional and subtraction taking places in the sound field. Y Axis represents Power Spectrum in dB. M ild M oderate Lab Results S im ulated Lab R esults S im ulated Pow er Spectrur 1 Density in dB20 . -20 -20 -40 -40, 4000 Frequency in Hz 6000 8000 2000 ) 4000 Frequency in Hz 8000 Severe Lab R esults S im ulated -20 -40 2000 4000 6000 6000 Frequency in Hz Figure 45: Contralateral Azimuth with FBC ON Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 Feedback Canceller Off The results from the physical system look very noisy. This is due to strong reflections coming from room windows and floor. Also the strong head shadow has their effects on the response. Y Axis represents Power Spectrum in dB. — Lab R e s u lts — S im u la te d — Lab R e s u lts — S im u la te d Power Spectrur 1 Density in dB 2DQ0 6D0D 2 O D 0 4DDD 6DDD 8DDQ 4DQD Frequency in Hz MILD HEA RING L O S S Frequency in Hz M O D E R A TE H EA RING L O S S Figure 46: Contralateral Azimuth with FBC OFF Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 Chapter 5 Conclusion The modeling study would prove helpful when trying to understand the summing of the natural and amplified inputs at the tympanic membrane. The model has been useful in testing the effects of various physical effects like reflection off from floor, from the window in the measuring booth and also the changes in the software codes that drive the TI DHP 100 sound processor. The model takes care of the system limitations including microphone noise floor and dynamic range. Hearing aid fitting rules have been understood and implemented to produce theoretically optimal outputs for individuals with different hearing loss conditions. The current work serves for helping achieve best outputs for subjects who are monaurally aided. A constant delay of 3ms (which is roughly 72 samples) has been added in all the impulse responses to avoid any problems such as non-causality in case of Ipsi azimuth. The model has a gain of approximately 30 dB and a dynamic range of 75 dB. The microphone used in this modeling study is FG 3329 with a sensitivity of 1.0V/0.1 Pa relative to -53 dB SPL. The output transducer used is Knowles Electronics’ Cl 8409.It has a sensitivity of about 93.2-mV/Pa.Upon tests with new transducer this number should be appropriately calculated and modified in the existing model. The TI Processor had been modeled in 2 ways: Line Mode where it simply passes the input thru output and a User Defined Mode which spans a range if variables such as FBC, Pitch Shifter, UCL’s etc. 5.1 Considerations for Future Work Human hearing and the ability to perceive the location of a sound source have long been accepted as a process requiring the use of two ears. This process is referred to as binaural hearing. Monaural hearing, as it relates to auditory localization, is limited to pinnae cues, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 which are functionally different from binaural cues. They are based on the quality of sound as it enters the ear canal. Most patients with bilateral hearing loss will benefit from binaural hearing aids. The benefits of binaural amplification are binaural summation, which is hearing threshold improvement listening with two ears instead of one, tuning out unwanted noise, and better sound localization. It would be definitely of importance if the quality of a sound entering the two ears is studied and may be manipulated through varying means. The convoluted structure of the pinna is such that sound waves, as they are gathered and funneled toward the ear canal, experience overlapping, cancellation, reverberation and reinforcement influences. These influences produce quality, or acoustical changes in the spectral frequency make-up of the sounds, and it is these changes that provide the monaural information necessary for determining where a sound is coming from. This model has been helpful in assisting evaluation of parameters and the exact amount of signal processing required for monaural hearing loss. Another possibility of future work that has not been yet explored is the losses in the open ear canal that might occur. These losses generally occur at low frequencies and might require higher amplifications at these frequencies. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 References 1. 2575 Reference Microphone Data Sheets, Larsen and Davis, 2000. 2. A Computational Model of Spatial Hearing, Keith Dana Martin, MIT, 1995. 3. C. Harris, Handbook of Acoustical Measurements and Noise Control, 1998 4. Cl Receiver Series Data Sheet, Knowles Electronics, 2003. 5. Corey I. Cheng, Gregory H. Wakefield, Audio Engineering Society, 1999. 6. FG Series Microphone Data Sheet, Knowles Electronics, 2001. 7. HTD Measurements, R.Peters, M.Burkhard, Knowles Electronics Inc, Illinois, 1963. 8. Hearing Aids, Harvey Dillon, Boomerang Press, 2001. 9. Hearing Aids and Room Acoustics, Boothroyd, The Hearing Journal, 2003. 10. Hearing Test Device Documentation, Shawn Gao, House Ear Institute, 2000. 11. M. Flynn, Scientific Basis for an Open Ear acoustic System, Hearing Review, 2003. 12. Middle Ear Implantable Hearing Aids, Mark Ross, Harry Levitt, 2001. 13. Moore, B.C.Ear & Hearing, 1996. 14. R. Fretz,P. Stypulkowski, Open ear canal hearing aid system, USP Assigned 1993. 15. R. Pogash, C. Williams, Occlusion Protocols and Strategies, Hearing Review, 2001. 16. S.Soli, maintaining directional hearing with hearing aids, USP, 1994. 17. S.Soli, Feedback suppression apparatus with adaptive filtering, USP 1995. 18. Sub band Feedback Active Noise cancellation, B Siravara, UT Dallas, 2002. 19. TMS320VC5402 Fixed Point DSP Technical Manual, Texas Instruments, 2000. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 Appendix Weiner filter program To obtain a Weiner filter response, supply the input and the output *.wav files. Select appropriate filter length and sampling rate (be it same as the sampling rate of the input sound file). The result is a weiner.flt file, which would be analyzed in MATLAB. MATLAB Stens 1. Read the .fit file in MATLAB as follows X=fltread(‘weiner. fit’) 2. To obtain a frequency response do freqz Syntax: freqz (x, 1, 1024, 24000) This would plot the transfer function and magnitude plot of the system configuration under consideration. W iener F ilter D esign Input! Output Write Bet to: to g ffc C:\ReceiveiY.wav C :\B eceim ft jcAwtenerloatst ■ i l f r r Use VC r Use VC & Append C Overwrite Strip the tint R e a lig n Rlter length: ITT length: Sampling rote: Use 0 samples of both sequences 0 samples [>G delay output; <0: delay input) samples Design T^Oglxj Browse... Browse... Brawn... 512 _ » aocb H r 4$XKl samples to calculate power spectra r Remove DC from sequences before design Calculate Figure of Merit using frequency range j0 to |8000 Ha rGraphtar— — — ......- Goodneweff* R i t e ? R e s p o n s e Figure 47: Weiner Filter Program Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 Hearing Test Device left Chanud A ile ro n Inpi* Output r ode r 20 d e r 40 dB ^ J & 60 d6 0 M ule •RigttChewieiAtteriuetorv iff)U Output f « J s d B r O dB G 20 dB r 40dB c 60de UsdContrafc...........................-............................... xLeO Monita R ight T atk Back TalkFotwafd 0 0 0 0 M ute M ute M ute M ote OutJXTrawtkJMrs............... * Speaker* i~ Headphone* < * BoneV&diw land dghtHP) f BoneVfciatwtandWtHP! T N ot RtS^t thennd Record Soutue r internal Lelt (S ’ Intern* R ight T External R ight r TdkFwwaid Enable ftj.-j 1 1 .1 , 4 I I , 1 1 1 I n (W W nanOSwcn I Figure 48: Hearing Test Device Select the following parameters • Check Output Transducer as “speakers” • Provide requisite input gain in the “Right Channel Attenuator” • Select Record source as “Internal Right” DSC .exe - f c L t f Plwlfe 1 8row *e.<. [ Aeond tie itncrw or sterec-leltk J Srawse... [ RecWdWaWewMght* | Browse.. J ~ Play Patawetsrs - Record Parameters-"--------- --------- -------\ ^ U ntil key or mouse I Loop Duration Isampfejf jlfkK jO u j T Ur« “Stop" button ........... I S«B*8nflwte|H*J; pOSO AUdnuaK ontdBk f~ ~ r Mom # Si«*o | S*gn»«(0«i8k f ~ • Mode— '-™ — .........................................^ r k»mmsHt segment T Play only Output Channels): C Recorded teft (S' Right <* Bolh < * Play and record 1 r Ovtttide voium a eonttol sattim* J j [ ...ptr... 1 Start “ 1 I i ............. ........................... 1 Figure 49: DSC.exe Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 • Supply the white noise file name under ‘Play File’ to be played and provide input and output file names. • The left channel is the Output and Right channel is Input (looped back to reach HTD same time as op so that no time delay exists between the op and ip). • Select the mode as ‘Play and Record’. • After setting up all parameters hit Start to begin recording. Simulink Files From W ave File W hiteN oiselO kH z.w avO ut - (10000H»M Ch/10b) From w a v e 0 061 3 MIC gain 2 DFFIR To W oitepaaea DF FIR A D igital Filters KBUAR FRONT T d WD)kjpaB«4 roam effects Uniform R andom N um ber Digital Filter ACOUSTIC SCALING © r f © To W o tep a ce FRONTsim DF FIR 5 T F I R To W oM p*oe3 0 .0 2 2 4 Digital Filter! OP Transducer Digital Filteiz Digital DSP Part Analog Part TI PROCESSOR Figure 50: Simulink Diagram Matlab Scripts Summation Study Script (Front only) %% Plotting PSD Functions for front Azimuth with WGN %% Total front responses at [2 2 2] Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 Axis ([0 8000 0 200]); Subplot (2, 2, 2), psd(squeeze(FRONT sim)* 10A (94/20), 1024,16000); Title ('COMBINED FRONT RESPONSE: SIMULATED '); Hold; %% WGN Power Spectrum Density along with front total resp at [2 2 2] Axis ([0 8000 0 200]); Subplot (2, 2, 2), psd(squeeze(WGN)* 10A (94/20), 1024,16000); % WGN Power Spectrum Density along with front UP resp at [2 2 1] Axis ([0 8000 00 200]); Subplot (2,2,1), psd(squeeze(WGN)* 10A (94/20), 1024,16000); %title ('PSD for White Noise '); Hold; %% front UP responses at [2 2 1] Axis ([0 8000 00 200]); Subplot (2, 2, 1), psd(squeeze(FRONT up) * 10A (94/20), 1024,16000); Title ('Upper part of the model'); Validation Study Script (for Front only) %% to plot the simulated results with lab results: model validation %% Load validapr!2FBCON.mat; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 %% FRONT MILD FBC ON Axis ([0 8000 -40 60]); fxl=wavread ('c: \FBC ON\fxl.wav'); fyl=wavread('c:\FBC ON\fyl.wav'); [psdopA,f] = psd((squeeze(fyl (48000:96000)*4.46)), 1024,16000,HAMMING( 1024)); [psdipA,f] = psd((squeeze(fx 1 (48000:96000))), 1024,16000,HAMMING(1024));axis([0 8000 -40 60]); Subplot (2, 2, 1) Plot (f, 10*logl0 (psdopA./psdipA),'-r');legend('Lab Results','Simulated'); Hold on; [psdopB,f] = psd((squeeze(Fsiml (48000:96000))), 1024,16000,HAMMING( 1024)); [psdipB,f] = psd((squeeze(ip 1 (48000:96000))), 1024,16000,HAMMING( 1024));axis([0 8000 -40 60]); Subplot (2, 2, 1) Plot (f, 10*logl0 (psdopB./psdipB),'-b');legend('Lab Results','Simulated'); %% FBC MOD FBC ON fx2=wavread('c:\FBC ON\fx2.wav'); fy2=wavread('c:\FBC ON\fy2.wav'); [psdopC,f] = psd((squeeze(fy2(48000:96000)* 14.46)), 1024,16000, HAMMING(1024)); [psdipC,f] = psd((squeeze(fx2(48000:96000))), 1024,16000,HAMMING( 1024));axis([0 8000 -40 60]); Subplot (2, 2, 2) Plot (f, 10*logl0 (psdopC./psdipC),'-r');legend('Lab ResultsVSimulated'); Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 O - Figure 51: TIDHP 100 Simulink Model Figure 52: FG 3329 Mic Response in front and back directions Back Contralateral Ipsiltaeral Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 Figure 53: FG 3329 Mic Response in contralateral and ipsilateral directions FG 3329 Microphone Fit ciro-Acoustic Supply Voltage RO Te............................. C-hoUVHC '.viral. SOliA iraximuir lj« .iltiyid in :h.ift . . . jlM B -'H H , ........................................................................... ZHluta.Bm 4 .4 typical DG.Outpu W m R a n * ......................... 0 /iuJW D C 3 bVDC ivncol RSrtr-liPtW S-dii .uni, Aihii.i.i'tei ZidBlvpi -ll ktupt.il id "re-t '.’M B nwrnm i i annum. A-wuiglneil tew&tte M anly l t . u ( i f t i i b j f at (Ci l . i k J inlet u i u s e s a pcisiuveyu.ng w lt.nc i - j dpp.:j at the outputtur initial. relative w i l i t i _______________________ Table 3: FG Series Specifications ‘d S 8 S 4 S 2 ......... 3S2S m - M i S • ■ ■ » H3-8161 — — - < «* ** - ■ -----------p* f * 0 0 * * * , * + m ....... .......... 3*” 100 $ 4 * 6 7 8 9 1003 2 3 Opwi ctraiu aiflfltwijy w&t 1.3VDC supply 4 5 8 7 Figure 53: FG 3329 Response as from data sheet Cl 8409 Receiver Performance Response/Sensitivity see curve, page 1 Sensitivity Range ±3dB @ 500Hz Total Harmonic Distortion 3 5% typical @ 500Hz. nominal drive Temperature Range operating: -17*C to 63°C storage: -40‘C to 63"C Acoustic Polarity 2-terminal: A positive going voltage at terminal 2 relative to terminal 1 causes an increase in pressure at the sound outlet 3-terminal: A positive going voltage at terminal 2 relative to the center-tap terminal or a negative going voltage at terminal 1 relative to the center tap ter minal. causes an increase in pressure at the sound outlet Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 Response C l I g y j (X . C Q t O Z U J to 135 01-274$ 130 125 120 115 110 105 100 1 0 0 FREQUENCY IN HERTZ H is% ponm measured using 8mm x m ini ♦ 28mm x 1.5mm + 25mm x 2mm ♦ 18mm x 3mm tubing mto 2cm* cavity Figure 55: Receiver Response 150 140 120 Output at t0% THD ® 500Hz 110 100 90 80 70 60 10000 100 FREQUENCY IN HERTZ Figure 56: Receiver Saturation Levels Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Head injury biomechanics: Quantification of head injury measures in rear-end motor vehicle collisions

PDF

Cellular kinetic models of the antiviral agent (R)-9-(2-phosphonylmethoxypropyl)adenine (PMPA)

PDF

A multimodal screen reader for the visually impaired

PDF

Effects of prenatal cocaine exposure in quantitative sleep measures in infants

PDF

Characteristics and properties of modified gelatin cross-linked with saline for tissue engineering applications

PDF

Contact pressures in the distal radioulnar joint as a function of radial malunion

PDF

A model of upper airway dynamics in obstructive sleep apnea syndrome

PDF

Estimation of optical fluorescent impulse response kernel in combined time and wavelength spaces

PDF

Comparisons of deconvolution algorithms in pharmacokinetic analysis

PDF

Cross-modal prepulse modification of ERP and startle

PDF

A model of cardiorespiratory autoregulation in obstructive sleep apnea

PDF

Evaluation of R.F. transmitters for optimized operation of muscle stimulating implants

PDF

A finite element model of the forefoot region of ankle foot orthoses fabricated with advanced composite materials

PDF

Three-dimensional functional mapping of the human visual cortex using magnetic resonance imaging

PDF

Functional water MR spectroscopy of stimulated visual cortex using single voxel

PDF

Design of a portable infrared spectrometer: application to the noninvasive measurement of glucose

PDF

Development of ceramic-to-metal package for BION microstimulator

PDF

Cardiorespiratory interactions in sleep apnea: A comprehensive model

PDF

Application of dynamic mixture model to fMRI

PDF

Finite element analysis of the effects of stem geometry, surface finish and cement viscoelasticity on debonding and subsidence of total hip prosthesis

Asset Metadata

Creator Chandwani, Neeta (author)

Core Title An open ear canal sound delivery system

School Graduate School

Degree Master of Science

Degree Program Biomedical Engineering

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag engineering, biomedical,Health Sciences, Audiology,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Khoo, Michael C.K. (committee member), Singh, Manbir (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-317133

Unique identifier UC11326965

Identifier 1422387.pdf (filename),usctheses-c16-317133 (legacy record id)

Legacy Identifier 1422387.pdf

Dmrecord 317133

Document Type Thesis

Rights Chandwani, Neeta

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA