課程名稱︰數位語音處理概論
課程性質︰選修
課程教師︰李琳山
開課學院：電資學院
開課系所︰電機、資工系
考試日期（年月日）︰2004.12.10
考試時限（分鐘）：120
是否需發放獎勵金：是
（如未明確表示，則不予發放）
試題 :
Digital Speech Processing
December 10 2004, 10:10-12:10
● OPEN EVERYTHING
● 除專有名詞可用英文外，所有文字說明一律以中文為限，未用中文者不予計分
● Total points: 120, Time Allocation: 1 point / minute
───────────────────────────────────────
1. (10)
(i) (5) What are voiced/unvoiced speech signals and their time-domain
waveform characteristics?
(ii) (5) What is the pitch in speech signals and how is it related to the
tones in Mandarin Chinese?
╴
2. (20) Given a HMM λ = (A, B, π), an observation sequence Ｏ = o_1 o_2 ...
o_t ... o_T and a state sequence q(上面加底線) = q_1 q_2 ... q_t ... q_T,
define
α_t(i) = Prob[o_1 o_2 ... o_t, q_t = i│λ]
β_t(i) = Prob[o_(t+1) o_(t+2) ... o_T│q_t = i, λ]
N
(i) (5) What is Σ α_t(i) β_t(i) ? Show your results.
i=1
(ii) (5) What is α_t(i) a_ij b_j(o_(t+1)) β_(t+1)(j)? Show your results.
(iii) (10) Formulate and describe the Viterbi algorithm to find the best
state sequence q*(上面加底線) = q_1* q_2* ... q_t* ... q_T* giving the
highest probability Prob[q*(上面加底線), Ｏ(上面加底線) │λ].
Explain how it works and why backtracking is necessary.
3. (10) Explain and describe what you know about "dialogue modeling and
management".
4. (10) Explain and describe what you know about "Text-to-speech Synthesis".
5. (10) Write down the procedures for LBG algorithm and discuss why and how it
is better than the K-means algorithm.
6. (10) Explain the detailed principles and process for Katz smoothing.
7. (10) What is the perplexity of a language source? What is the perplexity of
a language model with respect to a test corpus? How are they related to a
"virtual vacabulary"?
8. (10) Explain how the MAP principle can be used to find a word sequence.
Ｗ(上面加底線) = w_1 w_2 ... w_n given an observation sequence
Ｏ(上面加底線) = o_1 o_2 ... o_T, how the hidden Markov Model and language
model can be used, and which the likelihood function and the prior
probability are?
9. (30) Write down anything you learned about the following subjects that were
NOT mentioned in the class. Don't write anything mentioned in the class.
(i) (15) classification and regression tree (CART)
(ii) (15) search problem/algorithm for large vocabulary continuous speech
recognition