電腦有可能像音樂家一樣與人合奏嗎?本研究擬探討電腦與人合奏的可能性,研究虛擬音樂家或虛擬合奏的挑戰。我們計劃發展出一個電腦系統能夠跟人合奏 , 這個人可能是用傳統樂器,也可以是用MIDI電子樂器 , 未來也會探討肢體語言與電腦之間的合奏。 本研究將面對幾個技術的挑戰。首先我們需要有一個如同專業音樂家的音樂特徵探測器。將使用者彈奏的音符轉換成方便處理的音樂特徵,例如:和弦、調性……等等,再利用處理好的特徵跑即興演算法,產生即興樂曲與人合奏。系統需有一個好比演奏家心裡的音樂時脈(Music Clock),雖然獨立運作仍須不斷傾聽他人以隨時調整。我們希望發展出一個像真人合奏般的電腦與人合奏系統,所以即時音樂運算與互動是研究的關鍵,也是挑戰。
Can computers sing along or play music with human beings? This study aims to investigate the challenges and technical issues of virtual musicians or virtual ensemble. In essence, we want to design an intelligent music ensemble system to play along or sing along with human beings, who may use either traditional music instrument or digital MIDI-based music instrument. In the future, we will explore using body language to ensemble with computer. Our system needs to analyze the input music like a real musician who knows the scale, key, and chord very well, so it can harmonize melody or play the suitable melody on a chord. To present like a real musician, the system needs a ''Music Clock'' which still needs to be regulated by listening to other playing immediately. Just like a real ensemble situation, musicians need to listen to each other attentively to create a beautiful music. In the ensemble system, real-time interaction is the key in our research. Meanwhile, it is our challenge, too.