Project topics and materials
Get Complete Project Material File(s) Now! »
Vocal Signal Measurement Equipments
Actually, the microphone constitutes the most common tool to acquire the speech signals. However, the quality of the recorded signals is highly affected by the interference of the background noise. Since the speech signal and the noise have the same frequency’s band, it is very difficult to separate them and to perform a 100% extraction of the speech. This issue has been the researchers’ interest and has gained more and more attention. Many algorithms have been developed in order to eliminate or to reduce to a large extent the embedded noise and the majority have yielded good results. Besides, the research has been conducted to develop non acoustic means to acquire the speech. Any sensor which is able to collect the speech before it leaves the speaker’s lip/oral cavity is immune to the ambient environment noise [5, 6].
Non-acoustic measurement devices are usually robust and resistant to noise interference. In the past two decades, experiments using non-acoustic sensors have revealed that it is feasible to measure the glottal excitation and the articulator movements of the vocal cords in real-time as an acoustic speech signal is generated. The non-acoustic sensors can be classified into two categories: the physical instruments and the microwave devices. The physical instruments include mainly the ElectroGlottoGraph (EGG), the Tuned Electromagnetic Resonator Collar (TERC) sensor and the throat microphone. Among the microwave devices, the General Electromagnetic Micro-Power Sensor (GEMS) has played an important role in this area. It was used to measure the vocal folds’ vibrations during a speech. In addition, equipments such as the transnasal flexible endoscopy, the rigid endoscopy, the stroboscopy and the high speed video endoscopy have been designed to detect and to visualize the motion of the vocal folds [5-8].
ElectroGlottoGraph
The EGG (ElectroGlottoGraph) is a device that measures the Vocal Folds’ Contact Area (VFCA) [9]. That is, it measures the variations in the electrical resistance between two electrodes attached to the individual’s neck on each side of the thyroid cartilage (Figure 1.2). An electrical signal in the MHz range is sent through the neck of the subject. The VFCA is determined by observing the variations of the electrical impedance between the two electrodes when the vocal cords are in a vibration mode (individual speaking). The EGG provides a physiological measure of the fundamental frequency (Fo) of the vocal cords’ vibrations at the laryngeal source’s level. Compared to the acoustic signal, the EGG signal is much easier to analyze and to process [9-11].
The EGG has been implemented in many domains such as the speech recognition, the speaker authentication and medical applications. However, since the EGG provides a measure of the vocal cords’ contact, the sensor does not necessarily enable the observation of interesting phenomena during the open phase of the glottis. It can be noted here that the EGG is not an exact indicator of VFCA [9-11]. For example, during the transition to the open phase of the glottis, the mucus can “short out” the machine. That is, the glottis is closed when it is actually not the case i.e. the mucus bridging effect [12].
Glottal Electromagnetic Micro-Power Sensor
The GEMS (Glottal Electromagnetic Micro-power Sensor) is a non acoustic sensor that measures the opening and the closing of the glottis and the vocal cords’ movements based upon transmitting ElectroMagnetic (EM) waves into the glottal region. In other words, it measures the tissues’ movements in the human’s vocal tract during the phonation (Figure 1.4), including the vocal folds’ vibrations [5, 7, 9].
The old measurements with GEMS consist of strapping an antenna on the throat at the laryngeal notch or at other facial locations. This set up can make the subjects discomfort and sometimes may cause a skin irritation [5]. Subsequently, the radar technology has attracted a great interest in different domains, such as medical monitoring, speech and speaker recognition
Table of contents :
ACKNOWLEDGEMENTS
ABSTRACT
RÉSUMÉ
INTRODUCTION
Objective
Outline
CHAPTER 1: STATE OF ART
1.1 Introduction
1.2 Fundamentals of Voice Production
1.2.1 Breathing
1.2.2 Phonation
1.2.3 Resonance
1.3 Other Physical Factors
1.4 Vocal Signal Measurement Equipments
1.4.1 Electroglottograph
1.4.2 Tuned Electromagnetic Resonator Collar
1.4.3 Throat Microphone
1.4.4 Glottal Electromagnetic Micro-Power Sensor
1.4.5 Transnasal Flexible Endoscopy
1.4.5 Rigid Endoscopy
1.4.6 Stroboscopy
1.4.7 High Speed Video Endoscopy
1.5 Throat Microphone
1.5.1 Diagnostic
1.5.2 Speaker/Speech Recognition
1.6 Conclusion
CHAPTER 2: DEVELOPED APPROACH
2.1 Introduction
2.2 Developed Speaker Identification Approach
2.2.1 Signal Acquisition
2.2.1.1 Introduction
2.2.1.2 History
2.2.1.3 Domain of Application
2.2.1.4 Material’s characterization
2.2.1.5 Methodology
2.2.2 Short Time Fourier Transform
2.2.3 Normalization and Noise Removal
2.2.4 Features’ extraction
2.2.5 Database
2.2.6 Correlation
2.2.7 Principal Component Analysis (PCA)
2.3 Conclusion
CHAPTER 3: MODEL OF THE LAYERS OF THE HUMAN NECK
3.1 Introduction
3.2 System Model
3.2.1 Fluid Layer
3.2.2 Solid Layer
3.2.3 Fluid-Solid Interface
3.2.4 Reflection and Transmission Coefficients
3.2.5 Results
3.3 Experimental Evaluation
3.4 Conclusion
CHAPTER 4: RESULTS AND PERFORMANCE EVALUATION
4.1 Introduction
4.2 Method
4.3 Effect of the Window
4.4 Effect of the Time Step
4.5 Evaluation with Other techniques
4.5.1 Wigner-Ville Distribution
4.5.2 Choi-Williams Distribution
4.5.3 Results and Discussion
4.5.4 Quantitative Evaluation
4.6 Conclusion
CONCLUSION
Recommendations and Future Prospects
LIST OF REFERENCES
