INSTRU_1.pdf

Instrumentation in Speech- Language Pathology Professor Yaser S. Natour  types of measures Nominal A nominal measure is a group of categories that are not relative to each other but are different such as “male” versus “female” or “bird” versus “mammal.”  Nominal scales are referred...

Instrumentation in Speech- Language Pathology Professor Yaser S. Natour  types of measures Nominal A nominal measure is a group of categories that are not relative to each other but are different such as “male” versus “female” or “bird” versus “mammal.”  Nominal scales are referred to as qualitative groupings, that is, assigning certain attributes to items. An ordinal scale is where the attributes or categories are ranked relative to each other in a qualitative manner such as “mild,” “moderate,” and “severe” ratings of the degree of breathiness of a voice. These are discrete categories relative to each other, such as quantitative ratings on a scale.  Interval scale contains equal increments such that a value of 50 is greater than 40 to the same degree that 60 is greater than 50; such scales are often referred to as linear and continuous  Ratio scale is one that has a true zero value such as age, height, and weight. Values on a ratio scale are multiples of each other; for example, 40 is twice as large as 20  Accuracy of Measurement Precision includes several aspects of measurement such as calibration to read units in a system of units, that is, the reference standard such as millimeters or milligrams, and the accuracy of the measurement at different levels on the scale, such as the percent error from a reference standard at low and high levels.  Reliability of a measurement tool can affect the range of error that will occur when making a measure repeatedly on several persons, say a 5% error  Bias is if the measure usually differs from the reference in one direction.  Technical Issues Signal transducers (e.g., microphones, strain gauges, pressure sensors). A transducer is a device that changes one form of energy into another. The most common type of converted energy is an electrical signal  Active transducers do not require a power source, because they work on the principle of energy conversion to produce an electrical signal that is proportional to the physical property of interest. For example, a thermocouple is an active transducer used to measure temperature. It can be constructed from two legs of different metals, and temperature is measured at their junction point.  Passive transducers require a power source and typically work by generating an output signal of varying voltage or current. For example, a photocell is a passive transducer that is sensitive to the amount of light that falls on it but depends on a power source to generate an output signal.  a transducer is a device that can convert energy in any direction, both the microphone and the speaker are transducers. However, only the microphone is designed to function as a sensor in its capability to pick up signals that can control electrical or electronic circuits.  A sensor is a mechanical device sensitive to a particular physical property (such as heat, light, sound, pressure, magnetism, or motion) a mechanical device sensitive to a particular physical property (such as heat, light, sound, pressure, magnetism, or motion)  A signal is defined at only discrete instants of time. Can be discrete or continuous 1. The temperature taken at noon on successive days of the month of July (discrete) 2. The wind speed registered by an anemometer over the course of a day (continuous) 3. Heart rate determined at 10-min intervals during an exercise (discrete)  Analog signal is a continuous signal containing time-varying quantities. In telecommunications, an analog signal is one in which a base carrier’s alternating current frequency is modified in some way, for example, by amplifying the strength of the signal (amplitude modulation, or AM) or by varying the frequency (frequency modulation, FM).  A digital signal is discrete in both time and amplitude. A digital signal is discrete in time because the signal is sampled at regular intervals called the sampling rate and it is discrete in amplitude because this dimension is quantized , or represented in small increments (quanta). A signal can be periodic or Aperiodic  Signals of interest to SLPs Acoustic (audio) signals Aerodynamic signals Electrical and electromagnetic signals Mechanical signals Optical signals  signal amplification, signal filtering, anti- aliasing filtering ( filter used before a signal sampler to restrict the bandwidth of a signal), and digitization  signal amplification, signal filtering, anti- aliasing filtering ( filter used before a signal sampler to restrict the bandwidth of a signal), and digitization  Airflow, acoustic energy radiated in space, or electrical energy produced by nervous tissue, is picked up by a sensor or transducer. The signal so obtained is then sent to a processor, which can perform a variety of operations, depending on the purposes of measurement and analysis.  Storage devices also take several forms, some of the most common being the following: 1. Chart recorder, printing a hard copy of the data 2. Magnetic recorder, storing the data by means of magnetic fields 3. Photographic recorder, storing either still or motion pictures 4. Printer, producing a hard copy of computer- generated data in either alphanumeric or graphic format 5. Electronic memory, storing the data on devices such as disks or flash drives 6. Cloud storage, storing the data in multiple, virtual servers  Basics of Electricity Electricity involves a flow of charged particles. This flow is called current and is measured in amperes (often abbreviated as amps), but we rarely measure amps directly. In electrical wires, the charge is carried by (equivalently, the current is formed of ) moving electrons.  Personal Safety Whenever you work around significant current, remember to put one hand in your pocket. The reason for this simple safety rule is that you do not want current to go through your heart. Always wear shoes that insulate you from the ground, with thick rubber soles, for example, so that current cannot pass from hand to foot through the heart.  But hand-to-hand conduction can be more difficult to avoid if you do not put one hand in your pocket. The problem is often where you absent mindedly put your non-dominant hand.  Sometimes you can negligently place one (typically left) hand on some structure that is grounded, such as the metal casing around equipment, and then if you touch some stray or unintended current with your dominant hand, current could flow from hand to hand through the heart. Just remember to put your nondominant hand in your pocket whenever you are working around significant electricity.  Have someone with you at a distance when dealing with line current—the power in a wall outlet or greater. If you are ever dealing with more than 110 V alternating current (AC; like 220 V), have that extra person be near a phone to call for help and near a nonconducting pole such as a wooden broomstick to move you away from contact with current if you ever get shocked.  The current can temporarily overwhelm your own electrical neuromuscular signaling, and in some circumstances, you might be unable to move while in contact with current—not a good situation if you are by yourself.  Make sure your skin is dry. Damp skin has much lower electrical resistance than dry skin, so with the same voltage you will get more current through wet than dry skin.  Electricity Electricity is the movement of charged particles. Electrical current needs to go in a loop called a circuit. There can be active and passive components in a circuit. An active component, such as an amplifier, either requires or produces power.  passive component— such as a switch, resistor, or light bulb—needs no external power other than the current in the circuit in order to function. A circuit has a source of power, such as a battery or a generator, often a switch to open or close the circuit, and conductors allowing current to pass from the source through something of interest, like a light bulb, and then back to the source.  Basic anatomy of an instrumentation It is a device or system used to visualize the details of some phenomenon or process in an objective and perhaps even quantifiable way. ❖ Usually an instrument is “electronic”  Basic anatomy of an instrumentation Almost any instrument is built around three kinds of functional units: 1. The Input stage (Microphones, Transducer). 2. The Signal Conditioning stage (Amplification, Filtering, Computer digitization, Data calculation). 3. The Output stage ( Output displayer like; monitors, loudspeakers….).  Safety Electrical Safety Be Careful ✓ Before using any electrical instruments: 1. Be sure that the electrical outlets have a valid ground contact. 2. Don’t use an adapter to plug a three-prong plug into a two-prong outlet. 3. Never use any equipment that has worn wires. 4. Be with your patient on the safe side.  Safety Infection Control ✓Hepatitis ✓Herpes ✓Tuberculosis ✓AIDS ✓Bacterial infection ✓COVID 19 !  Safety We should minimize the risk of spreading infection. Handwashing/ Sterilizing/ Disinfectant. Using gloves/masks/ disposable tongue depressors.  Analog to digital conversion (ADC) ADC produces a binary output that can be read by a computer (graphed in Excel for example) proportional to the input voltage. A digital-to-analog converter (DAC) does the reverse, producing a voltage proportional to a digital input.  The sampling rate and the precision of the ADC or DAC. Sampling rate is how fast the conversions occur An important fact is to sample at least twice as fast as the highest frequency in your signal, or else you get a mistaken recording due to undersampling.  suppose the stock market went up one day and down the next and a not-so-clever trader only looked at the Dow Jones Average (DJA) every week. The trader would mistakenly believe the DJA went up and down once a week and would lose much opportunity for profit.  Amplification A preamplifier , for example, a condenser microphones also include preamplification using a battery to enhance signal changes prior to recording the voltage output. Differential amplifiers , amplifying the difference between electrical field changes picked up at two electrode locations slightly apart to sense muscle contractions beneath the skin in relation to a ground electrode. These are used in electrophysiology to measure muscle contractions  differential amplifier can measure very subtle changes in the electrical field as one muscle contracts and “ignores” potentially larger but irrelevant voltage changes (such as eye movements) that are picked up by both electrodes equally  Calibration Calibration of the signal must be completed for converting the output voltage signal into a reference standard measurement before digitization. By calibration, a voltage output is converted into a reference standard such as air pressure in kilopascals (kPa), airflow in milliliters per second, electromyographic recordings in microvolts, distance in millimeters, or current in milliamperes.  If no calibration is conducted, the voltage output can only be presented as a scale of arbitrary units (AU).  Calibration is determined when a known reference signal in the scale you are measuring is transduced and the voltage change that represents that signal change is measured An example from pressure measures is converting voltages that you measure as an output of metric of pressure in millimeters of cmH2O. Signal Filtering before Digitization If there is noise in frequencies outside the frequency range of the signal that you are interested in, then the signal may be filtered to include only the frequency range of interest. For example, most electromyographic (EMG) am plifiers for intramuscular recordings can be set to delete frequencies below 30 Hz and above 3000 Hz so that only those signal frequencies that reflect changes in motor unit f iring due to muscle contractions are transduced.  FOUR TYPES OF FILTERS 1. Low-Pass 2. High-Pass 3. Band-Pass 4. Band-Reject  1. Low-Pass Filter Passes energy below some f U: attenuates energy above f U Two parameters:  fU  Attenuation rate What is f ?  f = fU  2. High-Pass Filter Passes energy above some f L: attenuates energy below f L Two parameters; what are they?  fL  Attenuation rate What is f ?  f = f sig - f L  3. Band-Pass Filter Passes energy between some f L and f U: attenuates energy below f L and above f U All five parameters useful; what are they?  f c, f L, f U, f, and attenuation rate What is f?  f = f U - f L  Band-Pass Filter A band-pass filter is a combination of a low-pass & high-pass filter connected in series Ch6-55  4. Band-Reject Filter Rejects energy between some f L and f U A band-reject filter is a combination of a low- pass & high-pass filter connected in parallel  Digitization Digitization is the digital sampling of an analog voltage signal into a digital signal and is dependent upon the type of signal you are trying to capture for input to a computer for measurement.  Some signals are slow movement signals such as jaw position changes during speech or pressure changes in the pharynx during swallowing, which may have a frequency range from 0 to 10 Hz.  These signals have a small range and therefore the quantization is limited not requiring high resolution. In this case, the bit rate or number of “0” and “1” values needed to represent the range in values is relatively low.  Signals to be digitized also differ in their frequency components. Slow changes in voltages, such as occurs with slow movement like opening the jaw and closing the mouth or pressure changes in the pharynx during swallowing, have lower frequency components in comparison with relatively fast signals such as speech acoustic waveforms.  The frequency components of signals for voice, speech, and swallowing differ and require different sampling rates when digitizing signals.  Signals are usually captured either as alternating current or voltage (AC), whereas others are usually captured as direct current or voltages (DC). Examples of DC signals are changes in rib cage and abdominal expansion or retraction with respiration, which only change at frequencies less than 100 Hz as we breathe at rates of 16 to 20 breaths per minute.  lip and jaw positional changes during speech are also relatively slow within the range of 200 Hz. More rapid alternating (AC) signals oscillate around a zero voltage and usually need to be sampled at much higher rates; examples are EMG recordings of motor unit firings and acoustic speech recordings with frequency changes as high as 44,000 Hz if you wish to capture the highest range of human hearing, although 12,000 Hz is sufficient for good speech understanding.  Thus, speech and voice, a sampling rate greater than 26,000 samples per second is optimum to represent the rapid changes in voice. This rate is needed to be able to represent instabilities in the fundamental frequency to have enough samples to be able to see small rapid changes in voice  This rate is needed to be able to represent instabilities in the fundamental frequency to have enough samples to be able to see small rapid changes  In fact, many digital recording devices for speech use 44,100 samples per second; that is the rate of audio signals recorded on CDs. This sampling rate is used because the highest range of human hearing is about 22 kHz.  At the sampling rate of 44,000 samples per second in a voice with a fundamental frequency of 200 Hz, there will be close to 220 samples for each cycle of vibration, which may be needed to determine the cycle-to- cycle changes in periodicity in the fundamental frequency (referred to as jitter).  On the other hand, respiration is usually at 30 breaths per minute, and the frequency components are very slow, around 20 Hz. Therefore, the sampling rate for a movement signal such as respiration can be much slower, around 100–200 samples per second.  Nothing is gained by sampling at a higher rate than necessary and doing so increases the data storage requirements and computational time burden of the data analyses. Toward Objective Evaluation of Treatment Outcomes ▷ Clinicians, patients, and other stakeholders need objective measures to determine the success/failure of behavioral, medical, or surgical interventions. ▷ Treatment outcomes measures must be objective, valid, automated, and sensitive to heterogeneous voice qualities and severities. ▷ Historically, instrumental measures of vocal function have aimed to serve this purpose. Instrumental Measures in the Voice Lab ▷ Acoustic Recording and Analyses ▷ Aerodynamic Measurement ▷ Laryngeal Imaging ▷ Electroglottography (EGG) ▷ Laryngeal Electromyography (LEMG) Acoustic Recording and Analyses 73 Basics of Technical Instruments ▷ Signal Detection: ○ Microphone, camera, electrode, flow or pressure transducers. ▷ Signal Manipulation or Conditioning: ○ Filtering, amplification, digitization. ▷ Signal Reconversion: ○ Numerical form, visual display, speaker. Microphones ▷ Converting acoustic sound energy into mechanical energy, and then changing it again into electronic energy. ▷ This process is called “transduction”. Microphones are transducers. ▷ Microphones and associated signal recording characteristics (and processing) is specified according to: ○ Amplification ○ Amplifier Gain ○ Amplifier Linearity ○ Peak Clipping ○ Frequency Response ○ Filters Digital Signal Processing ▷ Analog to Digital (A/D) Conversion ▷ Conversion of analog signals (i.e., voice) into a digital format = digitization. ▷ Frequency and intensity of the signal must be converted accurately. ▷ A/D Conversion uses two key operations to assign numerical values that represent both time-varying frequency and amplitude features of the voice signal: ▷ Sampling ▷ Quantization Acoustic Measurements ▷ Can provide objective and noninvasive analysis of vocal function. ▷ Clinical utility of acoustic measures depends upon whether the acoustic measures: ▷ Can discriminate between normal and disordered voices. ▷ Correlate with auditory-perceptual judgments of voice quality and severity. ▷ Sufficiently stable to assess real change in performance across time. Acoustic Measurements ▷ Most acoustic measures are mathematical derivations of 5 common measures: ○ Fundamental Frequency ○ Intensity ○ Perturbation measures ○ Ratio of signal (i.e., harmonic) energy to noise ○ Spectral or Cepstral features Fundamental Frequency (Fo) ▷ Fo is the rate of vibration of the vocal folds and is expressed in Hertz (Hz) or cycles per second (cps). ▷ It is the reciprocal of the pitch period (T) i.e., the duration of a single vibratory cycle. So, Fo = 1/T. the higher the frequency the lesser the time required to complete one cycle ▷ “Pitch” is the perceptual correlate of Fo. Normative Fundamental Frequency Fundamental Frequency (Fo) ▷ Accurate Fo calculation depends upon valid pitch period (T) detection algorithms: ○ Peak-picking, ○ Zero-crossing, ○ Waveform matching. ▷ Clinical Measures related to Fundamental Frequency (Hz): ○ Mean Fo (connected speech vs. sustained vowels). ○ Fo Range (highest and Fo: Detecting fundamental frequency period in an acoustic lowest pitch a patient can waveform using the “peak-picking method” produce). Intensity ▷ Vocal intensity is the acoustic correlate of vocal loudness. ▷ Intensity is referenced to sound pressure level (SPL) and measured on a logarithmic decibel (dB) scale. ▷ Habitual intensity, and intensity range (max and min) are useful clinically. Intensity: Measuring amplitude height in an acoustic waveform. Intensity ▷ Intensity measures can be derived from a number of different instruments… ○ Sound Level Meters ○ Acoustic Analysis Programs, and in some cases… ○ Aerodynamic Measurement Devices Normative Intensity Frequency-Intensity Profiling ▷ Voice Range Profile ▷ Phonetogram ▷ Physiologic Frequency Range of Phonation ○ These tools provide a thorough description of a patient’s physiologic limits of frequency and intensity. ○ Especially useful for monitoring Voice range profile. Plot of vowel produced at minimum vocal range in professional voice and maximum intensity [decibel (dB) range on the vertical axis] across minimum and maximum frequency users. [hertz (Hz) and musical note range on the horizontal axis]. Acoustic Analysis of Voice ▷ Two major types 1. Time-based measures (e.g. perturbation measures such as jitter and shimmer) 2. Frequency-based measures (e.g. spectral and cepstral/peak) Perturbation Measures ▷ Cycle-to-cycle variability in a signal (typically measured from sustained vowel productions or “extracted” vowels from connected speech). ▷ Jitter = cycle-to-cycle variability in frequency (a.k.a. frequency perturbation, pitch perturbation). ▷ Shimmer = cycle-to-cycle variability in amplitude (a.k.a. amplitude perturbation). ▷ Calculation requires a quasi-periodic signal for reliable/valid perturbation analysis (i.e., Type I signal). Jitter ▷ Cycle-to-cycle variability in frequency ▷ Purpose: To measure frequency stability in sustained vowels. Jitter is not a meaningful measure for connected speech ▷ Procedure: Can be measured when recording sustained vowel(s) for MPT ▷ Norms: Adults: