QUESTION 9 Which of the following might include text to speech segments and/or animations, as well as natural audio and
Posted: Fri May 20, 2022 1:45 pm
QUESTION 9
Which of the following might include text to speech segments and/or animations, as well as natural audio and video?
A. MPEG-21
B. MPEG-4
C. MPEG-2
D. MPEG-7
QUESTION 10
What is the entropy, in bits/character, of the following message?
sally sells seashells
There are 21 characters; the counts are s:6, l: 6, e: 3, a:2, space: 2, h: 2, y: 1
A. 4 bits/character
B. 7 bits/character
C. 2.5 bits/character
D. 3.5 bits/character
QUESTION 11
The zonal mask used in jpeg achieves compression by
A. reducing the size of the image
B. throwing away some of the DCT coefficients
C. increasing the number of coefficients
D. reducing the brightness of the image
QUESTION 12
In class we considered entropy according to a zero-order model - in other words, given a text, we considered the probability of each character without reference to its context. What if we were to use a first order model, in which we consider the probability of each character given the character that appears before it? Or what if we were to use a second order model, in which we consider the probability of a character given the two characters before it? Which of the following statements is true?
A. We expect the entropy to be highest for the zero order model and lowest for the second order model
B. We expect the entropy would be the same for all three models
C. We expect the entropy to be lowest for the zero order model and highest-for the second order model
D. We expect the entropy to be highest for the first order model and lowest for the second order mode
QUESTION 13
A Haar transform can be used to compress a signal by
A. doing the transform multiple times
B. stopping the transform before carrying out all the steps
C. deleting detail coefficients that are lower than a threshold
D. reducing the size of the signal before starting the transform
QUESTION 14
It's easy to write a program to decode Huffman-coded messages because
A. Huffman codes use the same number of bits for all symbols so the boundaries are easy to find
B. Huffman codes are unique - there is only one possible code for the symbols making up any signal
C. Huffman codes possess the prefix property - the code for any given symbol can't be the beginning of the code for another symbol
D. there are only a few valid Huffman codes in existence
QUESTION 15
Which of the following could be a valid Huffman code? Select all that apply.
A. a: 00, b: 01, c: 111, d: 100, e: 1010, f: 1011, g: 1100, h: 1101
B. a: 11, b: 10, c: 000, d: 011, e: 0101, f: 0100, g: 0011, h: 0010
C. a: 10, b: 100, c: 010, d: 1010, e: 1100, f: 1110, g: 1001
D. a: 01, b: 00, c: 111, d: 110, e: 100, f: 1011, g: 1010
E. a: 00, b: 01, c: 010, d: 011, e: 0011, f: 0001, g: 0110
QUESTION 16
The "DC" component - coefficient[0][0] - in an image DCT corresponds to the
A. number of pixels in the block
B. number of operations needed to carry out the transform
C. average brightness over the block of pixels
D. average coefficient value
QUESTION 17
The jpeg2000 standard is
A. based on the wavelet transform, rather than the DCT
B. known for giving images a greenish tint
C. not as effective as the initial jpeg standard
D. older than the initial jpeg standard
QUESTION 18
Some speech processing algorithms use 10 ms (millisecond) blocks of samples. How many samples are in a 10 ms block if the sample rate is 44,100 Hz?
A. 44
B. 4
C. 441
D. 4441
QUESTION 19
Why does the jpeg standard organize its DCT coefficients in a zigzag order?
A. To make run length coding more efficient
B. It makes it easier for people to read the coefficients
C. To randomize the coefficients before further processing
D. It restores the DCT coefficients to their most natural order
QUESTION 20
A 60 dB tone at 1000 Hz is most likely to mask which of the following tones?
A.20 dB at 1200 Hz
B. 20 dB at 2000 Hz
C. 20 DB at 500 Hz
D. 20 dB at 800 Hz
QUESTION 21
Assume the following is a row of grayscale pixel values
[94 112 102 98 114 150 170 160 148 144 168 182 192 196 198 200]
How many steps are needed to complete the 1D Haar transform for this vector?
A. 4
B. 8
C. 16
D. 2
QUESTION 22
Assume you have carried out an analysis of an audio signal using an FFT with a frequency resolution of 100 Hz. Now you want to carry out another analysis with a frequency resolution of 50 Hz. How will you modify the block size?
A. Set the block size to 50 samples
B. Increase the block size to hold twice as many samples
C. Set the block size to the Nyquist frequency
D. Decrease the block size to hold half as many samples
QUESTION 23
Given an audio system with a 44.1 kHz sample rate (44,100 samples/sec), what block size (how many samples) will you use for a Fourier transform if you want a frequency resolution of 100 Hz?
A. 100
B. 50
C. 441
D. 882
QUESTION 24
Given the following Huffman code, what is the message for the following bit sequence?
x: 10; y: 11; z: 000; a: 001; b: 011; c: 0100; d: 0101
110011010010100011011
A. yaxdybda
B. yaxxdayb
C. dcbazyx
D. xyzabcd
QUESTION 25
Which of the following code segments will correctly add reverb to a sound, as described in class? Assume snd is a numpy array holding audio samples, the value of x is 1600, and the value of y is .2.
A. for i in range(x, len(snd)):
snd = (1-y) * snd + y * snd[i-x]
B. for i in range(x, len(snd)):
snd = y * snd + (1-y) * snd[i-x]
C. for i in range(x, len(snd)):
snd += x * y
D. for i in range(len(snd)):
snd = snd + snd[i-x] * x + y
Which of the following might include text to speech segments and/or animations, as well as natural audio and video?
A. MPEG-21
B. MPEG-4
C. MPEG-2
D. MPEG-7
QUESTION 10
What is the entropy, in bits/character, of the following message?
sally sells seashells
There are 21 characters; the counts are s:6, l: 6, e: 3, a:2, space: 2, h: 2, y: 1
A. 4 bits/character
B. 7 bits/character
C. 2.5 bits/character
D. 3.5 bits/character
QUESTION 11
The zonal mask used in jpeg achieves compression by
A. reducing the size of the image
B. throwing away some of the DCT coefficients
C. increasing the number of coefficients
D. reducing the brightness of the image
QUESTION 12
In class we considered entropy according to a zero-order model - in other words, given a text, we considered the probability of each character without reference to its context. What if we were to use a first order model, in which we consider the probability of each character given the character that appears before it? Or what if we were to use a second order model, in which we consider the probability of a character given the two characters before it? Which of the following statements is true?
A. We expect the entropy to be highest for the zero order model and lowest for the second order model
B. We expect the entropy would be the same for all three models
C. We expect the entropy to be lowest for the zero order model and highest-for the second order model
D. We expect the entropy to be highest for the first order model and lowest for the second order mode
QUESTION 13
A Haar transform can be used to compress a signal by
A. doing the transform multiple times
B. stopping the transform before carrying out all the steps
C. deleting detail coefficients that are lower than a threshold
D. reducing the size of the signal before starting the transform
QUESTION 14
It's easy to write a program to decode Huffman-coded messages because
A. Huffman codes use the same number of bits for all symbols so the boundaries are easy to find
B. Huffman codes are unique - there is only one possible code for the symbols making up any signal
C. Huffman codes possess the prefix property - the code for any given symbol can't be the beginning of the code for another symbol
D. there are only a few valid Huffman codes in existence
QUESTION 15
Which of the following could be a valid Huffman code? Select all that apply.
A. a: 00, b: 01, c: 111, d: 100, e: 1010, f: 1011, g: 1100, h: 1101
B. a: 11, b: 10, c: 000, d: 011, e: 0101, f: 0100, g: 0011, h: 0010
C. a: 10, b: 100, c: 010, d: 1010, e: 1100, f: 1110, g: 1001
D. a: 01, b: 00, c: 111, d: 110, e: 100, f: 1011, g: 1010
E. a: 00, b: 01, c: 010, d: 011, e: 0011, f: 0001, g: 0110
QUESTION 16
The "DC" component - coefficient[0][0] - in an image DCT corresponds to the
A. number of pixels in the block
B. number of operations needed to carry out the transform
C. average brightness over the block of pixels
D. average coefficient value
QUESTION 17
The jpeg2000 standard is
A. based on the wavelet transform, rather than the DCT
B. known for giving images a greenish tint
C. not as effective as the initial jpeg standard
D. older than the initial jpeg standard
QUESTION 18
Some speech processing algorithms use 10 ms (millisecond) blocks of samples. How many samples are in a 10 ms block if the sample rate is 44,100 Hz?
A. 44
B. 4
C. 441
D. 4441
QUESTION 19
Why does the jpeg standard organize its DCT coefficients in a zigzag order?
A. To make run length coding more efficient
B. It makes it easier for people to read the coefficients
C. To randomize the coefficients before further processing
D. It restores the DCT coefficients to their most natural order
QUESTION 20
A 60 dB tone at 1000 Hz is most likely to mask which of the following tones?
A.20 dB at 1200 Hz
B. 20 dB at 2000 Hz
C. 20 DB at 500 Hz
D. 20 dB at 800 Hz
QUESTION 21
Assume the following is a row of grayscale pixel values
[94 112 102 98 114 150 170 160 148 144 168 182 192 196 198 200]
How many steps are needed to complete the 1D Haar transform for this vector?
A. 4
B. 8
C. 16
D. 2
QUESTION 22
Assume you have carried out an analysis of an audio signal using an FFT with a frequency resolution of 100 Hz. Now you want to carry out another analysis with a frequency resolution of 50 Hz. How will you modify the block size?
A. Set the block size to 50 samples
B. Increase the block size to hold twice as many samples
C. Set the block size to the Nyquist frequency
D. Decrease the block size to hold half as many samples
QUESTION 23
Given an audio system with a 44.1 kHz sample rate (44,100 samples/sec), what block size (how many samples) will you use for a Fourier transform if you want a frequency resolution of 100 Hz?
A. 100
B. 50
C. 441
D. 882
QUESTION 24
Given the following Huffman code, what is the message for the following bit sequence?
x: 10; y: 11; z: 000; a: 001; b: 011; c: 0100; d: 0101
110011010010100011011
A. yaxdybda
B. yaxxdayb
C. dcbazyx
D. xyzabcd
QUESTION 25
Which of the following code segments will correctly add reverb to a sound, as described in class? Assume snd is a numpy array holding audio samples, the value of x is 1600, and the value of y is .2.
A. for i in range(x, len(snd)):
snd = (1-y) * snd + y * snd[i-x]
B. for i in range(x, len(snd)):
snd = y * snd + (1-y) * snd[i-x]
C. for i in range(x, len(snd)):
snd += x * y
D. for i in range(len(snd)):
snd = snd + snd[i-x] * x + y