Recurrent Neural Network Synthesis (特集 音楽情報科学)

スポンサーリンク

概要

• 論文の詳細を見る

• This paper introduces a new technique of sound synthesis RNNS (Recurrent Neural Network Synthesis) by recurrent neural network. As a model of individual neuron, a continuous-time, continuous-variable neuron model is adopted. RNNS has two variations of sound synthesis : One variation of sound synthesis is "hand-tuning model", which is composed of relatively small network and is suitable for realtime sound synthesis. And another variation of sound synthesis is "resynthesis model", which is made of relatively large numbers of neurons. The latter is not only capable of learning of the original sound of the musical instruments, but is also capable of producing a new sound by modifying parameters of the neurons after learning process was finished.

• 一般社団法人情報処理学会の論文
• 2002-02-15

著者

• Ohya Ken'ichi

  Nagano National College Of Technology

関連論文

• Recurrent Neural Network Synthesis (特集 音楽情報科学)

スポンサーリンク