Automatic Composition of Music by LSTM

摘要

LSTM is used in different fields, such as Machine Translation, Time Series Prediction and so on. Machine Translation with LSTM uses Embedding vector for word and computes loss of two kinds of languages. And decreasing the loss of the Model and fixing it, which can translate one language to another one. Music and its lyrics are governed by rules which are declarative and non-deterministic. In this paper, compositing will be addressed with LSTM which can automatic analysis the rules of the music scores and lyrics. Automatic Composing can be looked the same as the Machine Translation. The Music score is a kind of language and the lyrics is another kind of language, which can use the Translation Model to fix it, and then to implement composing tasks.