COMD 5070 exam 3 study guide1.what is science? [4 features of the scientific method]empirical: collect and interpret data.deterministic: physical laws apply to principals. not random. generalizations from data to extend on to othersituationspredictive: learn cause and effect relationshipsparsimony: describe things in fairly economical terms2.semitones [how many semitones in one octave, two octaves, how many Hz in an octave, etc.]octave: doubling or halving of a frequency. 1 octave:12 semitones., semitone=1/12 octave - Each semitoneis a nonlinear step as you go from low to high- Each step up is bigger than the last one, about 5.9%higherfrequency than the one before it. Octave: mathematical relationship btwn freqs; doubling /half offrequency 400= 800 (up) & 200 (down) -12 semitone= 1 octave3.sampling rate [how does it relate to playback quality, frequencies saved in a recording, Nyquist,aliasing, etc.?]the frequency with which numbers are stored. recordings are a series of discrete snapshots in rapid successionwhich represents the amplitude at one given moment in time. Hz. Higher sampling rate with more samples willgive you a higher fidelity recording. Nyquist: half the sample rate. represents the highest frequency that therecording can accurately reproduce when you play it back. the higher the sample rate the better the quality, storesmore digits per second. the lower: fits more minutes.4.filter types [what do high or low pass or band pass or band reject filters do?]lowpass filter: prevents aliasing, prevents higher frequencies from reaching the digitizing system so they won’tcontaminate the lower frequency recordings by being represented as an alias, deletes frequencies above theNyquist frequency.high:allows high frequencies to pass through, removes or attenuates lower frequencies.band:removes low and high and allows a band in the middle to pass.5.time vs. frequency domain displays [what’s on the axes, what do the displays show us?]frequency domain:take the time domain wave form and do a Fourier transform to get a spectrum display.time domain:frequency and do a Fourier transform to get a spectrum display. you can see changes in therelative strength of the harmonics. If you raise or lower your pitch you see the harmonics spreading apartor coming back together. noise: all sorts of frequencies present with various different phase relationshipsto one another that are approx. equal in amplitude.6.types of spectra: line, FFT, LPC [they reveal different features of speech; what is each one best suitedfor?]A pure tone has a single vertical line on a spectrum.FFTfast fourier transform: the range of harmonics that arepresent in a complex sound. useful for revealing features of a sound source but not about the vocal tract filter.LPC: linear predictive coding. shows a spectral envelope. will not show individual harmonics or details of asound source but will reveal what the vocal tract filter is doing, how the vocal tract shapes the sound given to itby the larynx.7.3D spectrogram features [what are the three dimensions?]Spectrogram: way to represent sound graphically. time-left to right on the x-axis. frequency going up on they axis. the darkness of the trace is reflective of the sound at any given frequency and at any time on thedisplay.Preview Mode
This document has 14 pages. Sign in to access the full document!