1 |
一個新的庶民音樂創作經驗:智慧型手機上的配樂應用程式 / A New Experience of Music Creation for Plebeian: Musical Accompaniment Apps on Smartphone戴張戎, Tai, Chang Jung Unknown Date (has links)
長久以來音樂於人們生活中扮演著極為重要角色,在大多數人的成長過程裡或多或少皆有令其印象深刻之旋律。然而這些旋律常由專業人士所創作,對於未接受過專業訓練的民眾而言,若欲創作自己專屬之音樂難度甚高,而此目標也變得遙不可及。
為解決上述問題,降低音樂自行創作門檻,本研究以行動裝置之使用環境不受限及直覺性觸控介面兩大特性為運行環境,設計Android系統上之音樂創作軟體,協助未受過音樂專業訓練的庶民透過音樂主旋律並搭配適合的和弦配樂達成自行創作音樂之目標。本創作音樂軟體利用行動裝置提供「繪畫旋律曲線」與「字詞輸入」兩種輸入方式,將使用者繪畫的旋律曲線轉換為一段音樂主旋律,進行調性判斷、修正主旋律組成音並利用音樂動機樣式變化加以使主旋律更為豐富,輔以隱藏式馬可夫模型產生適切之和弦序列。最後將主旋律聲波與其產生的和弦聲波以混音的結果呈現給予使用者。
為評估本創作軟體是否符合使用者需求,以實驗觀察法邀請38位受試者進行軟體操作與評估。分析結果顯示,近75%的受試者認為由音樂創作軟體所產生之主旋律與和弦彼此搭配良好且符合其音樂動機。在介面易用性評估方面,結果顯示有近90%受測者認為本研究所提出的音樂創作軟體具有簡單易用之特性且能夠協助其降低創作音樂之門檻。簡單且易用的音樂創作軟體在實務上之重要性不言可喻,不但可使非專業使用者達成自我創作音樂之夢想,更可讓其沉浸於音樂創作成就感之中。 / Music plays as an essential role in human life and it affects the listeners on a certain extent. However, a pleasing music is the production of musicians and is difficult to be created by novices without musical specialty. To lower the entry point of music creation, this thesis design and develop a music accompaniment system on Android with the characteristics of intuitive input and ubiquity for novices without professional music background.
The developed system consists of the following modules, main melody preprocessing (key determination and melody modification), music similarity retrieval, main melody post processing (music motif variance), chord accompaniment (Hidden Markov Model and mixing main melody and chord melody) and text processing (tone determination and pitch finding) to automatically match the accordance between melodies and chords that are inputted by patting or word.
Thirty-eight participants were invited for system evaluation using the observational experiment. Nearly 75% of participants perceived that the melody and chord matching fits their musical motivations, while 90% stated that they can rely on the system to easily produce desirable music. Our findings contribute to the essence of music creation that the system provides a simplified interface for novice being immersed in music accomplishments, similar to that of professional musicians.
|
2 |
有記憶性信用價差期間結構模型李弘道 Unknown Date (has links)
本文建立了當違約機率及回收率為隨機變動,同時信用等級移動有記憶性,且回收率和無風險利率期間結構相關之信用風險價差期間結構模型。並評價信用價差選擇權及有對手違約風險普通選擇權之價值。
此模型產生的信用價差有更多的變化性,將可描述:信用價差的隨機波動行為,且即使信用等級沒變,價差仍可能發生改變;信用價差與無風險利率期間結構有相關性;公司特定或證券特定的價差及其變動行為;處於等級上升或下降趨勢公司債券之殖利率曲線,能更準確配適有風險債券的價格等實際現象。
並可應用至有對手違約風險之商品及多種信用衍生性商品等的評價與避險,且可進行風險管理方面的應用。
關鍵詞:信用風險;信用風險價差;馬可夫模型;信用衍生性商品 / In this thesis we develop a credit migration model with memory for the term structure of credit risk spreads. Our model incorporates stochastic default probability, stochastic recovery rate, and the correlation between the recovery rate and the term structure of risk-free interest rates. We derive valuation formulae for a credit spread option and a plain vanilla option with counterparty risk.
This model provides greater variability in credit spreads, and it has properties in line with what have been observed in practice: (1) credit spreads show diffusion-like behavior even though the credit rating of the firm has not changed; (2) the model injects correlation between spreads and the term structure of interest rates; (3) the model enables firm-specific and security-specific variability of spreads to be accommodated; and (4) the model enables us to estimate the yield curves corresponding to the positive and negative trends of credit ratings and match the observed risky bond prices more precisely.
This model is useful for pricing and hedging OTC derivatives with counterparty risk, for pricing and hedging credit derivatives, and for risk management.
Key Words: Credit Risk, Credit Risk Spread, Markov Model, Credit Derivative.
|
3 |
利用機器學習技術找出眼動軌跡與情緒之間的關聯性潘威翰 Unknown Date (has links)
目前偵測一般人情緒的方式大部分在研究人的行為,例如:臉部表情,以及分析人體的各項生理數值,例如:心跳、體溫以及呼吸頻率。然而這些研究只單純探討人的外在行為或生理訊號在不同情緒下的變化,而人的眼睛包含外在行為跟生理訊號,本研究將探討不同情緒下眼睛有什麼特別的反應。
我們先制訂一套實驗流程,在流程中我們以不一樣的情緒圖片給予受測者刺激,然後記錄受測者的眼動反應,並且讓受測者回報自己的情緒狀態。本研究也記錄受測者在情緒刺激下的眼動反應,並將眼動之反應轉換成序列資料,再針對不同情緒下的序列建立隱藏馬可夫模型(Hidden Markov Models:HMM)。希望藉著情緒模型,從眼動行為中偵測受刺激者處於何種情緒狀態。
本研究發現人在看圖時會依據對圖片內容的好惡,產生有意義的眼動反應。我們利用相對應的眼動反應建立情緒辨識系統,在辨識三種情緒時,辨識率能夠達到六成。
|
Page generated in 0.0254 seconds