O (8)

Question 1

What is the linear prediction model in spectral envelope analysis?

Answer 1

Auto-regressive (AR) model: H (2) = 1 / p-1 * Σ (from k=0 to p-1) where a0 = 1 and āk = ak/G

This model represents the frequency response in terms of past values.

Question 2

What is the cepstral analysis model in spectral envelope analysis?

Answer 2

Exponential model: H (z) = exp(Σ (from k=0) of b_k)

This model involves taking the reciprocal and changing the exponent’s sign.

Question 3

What does PLP stand for in spectral envelope analysis?

Answer 3

Perceptual Linear Prediction

PLP features are used to model the human auditory perception.

Question 4

In the context of spectral envelope, what does AR stand for?

Answer 4

Auto-regressive

This model predicts future values based on past values.

Question 5

What is the significance of model order in LP and CC features?

Answer 5

Determines the complexity and accuracy of the model

Higher model orders can capture more details but may lead to overfitting.

Question 6

True or false: The cepstral analysis model uses linear prediction.

Answer 6

FALSE

Cepstral analysis employs an exponential model rather than a linear prediction model.

Question 7

Fill in the blank: The frequency response in linear prediction is represented as H (2) = _______.

Answer 7

1 / p-1 * Σ (from k=0 to p-1)

This formula defines the relationship between past and current values in the model.