COMD 5070 CORE EXAM TEST 2026 QUESTIONS AND
SOLUTIONS RATED A+
✔✔Filter types - ✔✔high pass - allows high frequencies through, keeps out lower
frequencies
low pass - allows low frequencies through, keeps out higher frequencies
band pass - allows a band of frequencies through, removing frequencies that are higher
and lower
band reject - holds back specified band of frequencies and allows those that are higher
and lower to pass through
✔✔Reasons for using band-pass filter - ✔✔-used in the early days of speech analysis
-the output of various filters were measured
-the energy of each frequency level is shown
-a crude
✔✔Reasons for using band reject filter - ✔✔-remove unwanted noise
✔✔Adjustable band-pass filter - ✔✔-this filter allows you to look at different frequencies
✔✔Fourier theorum - ✔✔-All periodic sounds are made of a combination of sine waves
with varying amplitudes, phase angles and frequencies.
-Complex waves can be broken down into individual components.
✔✔Joseph Fourier - ✔✔1768-1830, a french mathematician & physicist.
✔✔Fourier transform - ✔✔-it is possible to analyze a time domain waveform to create a
spectrum which shows the individual components. This transforms waveform from time
domain to frequency domain.
✔✔Time domain data - ✔✔-a waveform represents sound directly
-air pressure changes over time
✔✔Frequency domain data - ✔✔-a line spectrum shows the frequency components of a
periodic sound
✔✔Periodic signal spectrum - ✔✔-has harmonics that are multiples of the fundamental.
Lines represent the harmonic frequencies.
-There is nothing between the lines.
✔✔Spectra (aperiodicity) - ✔✔-sine wave - single line on a spectrum
-complex periodic signals: multiple lines
✔✔Noise - ✔✔-all frequencies
, -equal amplitude
-random phase
-different kinds of noise: white, pink, brown, etc.
✔✔Real voice signals - ✔✔-the voice source is nearly but not truly periodic
-spectrum does not have pure lines
-spectrum has peaks
-there is some spread of energy around the fundamental and harmonics
✔✔FFT spectrum - ✔✔-Fast Fourier Transform
-clearly shows harmonic energy
-each peak is a harmonic
-formants are less clear
-more revealing of source vs. vocal tract filter
✔✔LPC - ✔✔-Linear Predictive Coding
-shows spectral envelope
-does not show harmonics
-good at revealing formants
-more revealing of filter, not source
✔✔Line spectrum - ✔✔-is a snapshot in time
-spectrogram shows speech over time
-spectra are lined up sequentially
-single slices are put together
✔✔Speech spectrogram - ✔✔x-axis is time
y-axis is frequency
darkness indicates intensity
✔✔spectrogram parameters - ✔✔-y-axis (fréquence) limited to Nyquist
-need to sample at high enough rate
-the display can be adjusted downward
-can't be adjusted beyond Nyquist frequency because you can't recover info that isn't
there
✔✔Analysis bandwidth - ✔✔-'wide band' spectrogram gives clear temporal detail
*frequency resolution is poor
-'narrow band' spectrogram gives clear frequency detail
*time resolution is poor
✔✔Vowels - ✔✔-vocal tract shape can be held constant
-time-invariant: a vowel can be prolonged
-one point in time looks like any other
-contrast with consonants: many are brief
SOLUTIONS RATED A+
✔✔Filter types - ✔✔high pass - allows high frequencies through, keeps out lower
frequencies
low pass - allows low frequencies through, keeps out higher frequencies
band pass - allows a band of frequencies through, removing frequencies that are higher
and lower
band reject - holds back specified band of frequencies and allows those that are higher
and lower to pass through
✔✔Reasons for using band-pass filter - ✔✔-used in the early days of speech analysis
-the output of various filters were measured
-the energy of each frequency level is shown
-a crude
✔✔Reasons for using band reject filter - ✔✔-remove unwanted noise
✔✔Adjustable band-pass filter - ✔✔-this filter allows you to look at different frequencies
✔✔Fourier theorum - ✔✔-All periodic sounds are made of a combination of sine waves
with varying amplitudes, phase angles and frequencies.
-Complex waves can be broken down into individual components.
✔✔Joseph Fourier - ✔✔1768-1830, a french mathematician & physicist.
✔✔Fourier transform - ✔✔-it is possible to analyze a time domain waveform to create a
spectrum which shows the individual components. This transforms waveform from time
domain to frequency domain.
✔✔Time domain data - ✔✔-a waveform represents sound directly
-air pressure changes over time
✔✔Frequency domain data - ✔✔-a line spectrum shows the frequency components of a
periodic sound
✔✔Periodic signal spectrum - ✔✔-has harmonics that are multiples of the fundamental.
Lines represent the harmonic frequencies.
-There is nothing between the lines.
✔✔Spectra (aperiodicity) - ✔✔-sine wave - single line on a spectrum
-complex periodic signals: multiple lines
✔✔Noise - ✔✔-all frequencies
, -equal amplitude
-random phase
-different kinds of noise: white, pink, brown, etc.
✔✔Real voice signals - ✔✔-the voice source is nearly but not truly periodic
-spectrum does not have pure lines
-spectrum has peaks
-there is some spread of energy around the fundamental and harmonics
✔✔FFT spectrum - ✔✔-Fast Fourier Transform
-clearly shows harmonic energy
-each peak is a harmonic
-formants are less clear
-more revealing of source vs. vocal tract filter
✔✔LPC - ✔✔-Linear Predictive Coding
-shows spectral envelope
-does not show harmonics
-good at revealing formants
-more revealing of filter, not source
✔✔Line spectrum - ✔✔-is a snapshot in time
-spectrogram shows speech over time
-spectra are lined up sequentially
-single slices are put together
✔✔Speech spectrogram - ✔✔x-axis is time
y-axis is frequency
darkness indicates intensity
✔✔spectrogram parameters - ✔✔-y-axis (fréquence) limited to Nyquist
-need to sample at high enough rate
-the display can be adjusted downward
-can't be adjusted beyond Nyquist frequency because you can't recover info that isn't
there
✔✔Analysis bandwidth - ✔✔-'wide band' spectrogram gives clear temporal detail
*frequency resolution is poor
-'narrow band' spectrogram gives clear frequency detail
*time resolution is poor
✔✔Vowels - ✔✔-vocal tract shape can be held constant
-time-invariant: a vowel can be prolonged
-one point in time looks like any other
-contrast with consonants: many are brief
