lec 5

Question 1

waveform display and editing + playback are analyzed within the ___ domain

Answer 1

time

Question 2

spectrums + formant analysis are analyzed within the ___ doman

Answer 2

frequency

Question 3

fundamental frequency analysis is analyzed within the ___ domain

Answer 3

time and frequency

Question 4

on a spectrogram, what is the relationship between the darkness of the bands and the amplitude?

Answer 4

• darker band = higher amplitude
• lighter band = lower amplitude

Question 5

what are the y and x axis of a waveform display vs spectrogram?

Answer 5

• waveform display: y = amplitude, x = time
• spectrogram: y = frequency, x = time (and bands = amplitude)

Question 6

what is sampling?

Answer 6

continuous signal (analog) being converted to discrete signal (digital) by measuring continuous signal at regular intervals

Question 7

what is a sampling rate?

Answer 7

number of measurements per second

Question 8

what is nyquist theorem?

Answer 8

• the highest frequency that can be adequately sampled = 0.5 * SR
• EXAMPLE: if your SR = 1000, the nyquist frequency = 500

Question 9

what happens to frequencies above the nyquist frequency? (in other words: when signal is undersampled)

Answer 9

• aliasing effect: frequencies above nyquist frequency “fold over” to lower frequency domain

Question 10

the SR should be ___ the value of the highest frequency.

Answer 10

twice!

Question 11

what is quantization?

Answer 11

converting a continuous range of values into a finite set of discrete levels

Question 12

1 byte = __ bits

Answer 12

8

Question 13

what’s the math behind 8 bits, 12 bits, and 16 bits?

Answer 13

• 8 bits = 2^8 = 256 levels
• 12 bits = 2^12 = 4096 levels
• 16 bits = 2^16 = 65,536 levels

Question 14

why are more bits better?

Answer 14

more bits = better representation of true amplitude changes

Question 15

if you have a 10V signal, what would be the mV/level for an 8 bit system vs 16 bit system?

Answer 15

• 8 bit system: 10/256 = 39mV/level
• 16 bit system: 10/65,536 = 0.15mV/level

Question 16

what is SNR ratio?

Answer 16

• signal to noise ratio
• SNR (dB) = 6.02 * Bit Depth + 1.76 (formula provided)
• just know that bit depth = how many bits and bigger number = less mushed signals

Question 17

what is pre-amplification or pre-emphasis?

Answer 17

a filter technique that enhances specific frequency regions (especially high freqs)

Question 18

define:

a) high pass fliter
b) low pass filter
c) band pass filter
d) band reject filter

Answer 18

a) allows high freqs thru
b) allows low freqs thru
c) allows only certain freqs thru
d) blocks bands of freqs

Question 19

why do people sound different on your cell phone than in person?

Answer 19

band pass filtering

Question 20

in digital filters, what is the bandwidth defined by?

Answer 20

window size

Question 21

what happens to temporal and frequency differences in large windows vs small windows?

Answer 21

• large window: temporal smearing but detects small freq differences
• small window: freq smearing but detects small temporal differences

Question 22

what kind of filters are large vs small windows associated with?

Answer 22

• large window = narrow-band filter
• small window = wide-band filter

Question 23

what is frequency resolution?

Answer 23

• ability to distinguish between different frequencies in a signal (smaller = better)
• Fres = SR / window size (in # of samples)
• divide 1000msec by SR to get # of samples value, then divide window size by that value and plug into formula