Design And Implementation Of System Of Lyric Generation Based On Neural Networks

With the promotion of material conditions,people have more pursuit for spirit.Music is a very important part in people's life,and music is also an inherent requirement for human.Lyric play a crucial part in songwriting.The quality of the lyric determines the quality of the song.In recent years,the neural network has been developing continuously,which has been used in image,speech and natural language processing.Especially in the field of natural language processing,the emergence of neural network not only provides an alternative route for academia but also provides a feasible technical route for industry.Therefore,this paper combines the relevant characteristics of lyrics and the advantages of neural network in natural language processing,and puts forward a method to generate lyrics and implement it.At first,this paper introduces some difficult points of lyrics generation and the advantages and disadvantages of different methods.In the case of Chinese,we study different methods' effect used for the lyrics generation.This paper mainly includes the process related content such as rhyme,lyrics length,lyrics context correlation,etc.The hardest part in lyric generating is that the generated lyrics' context needs to be relevant.This is hard for traditional statistical methods.In the field of natural language processing,the emergence of RNNs is of great significance.RNNs has the effect of dealing with indefinite long content,and it can also retain as much information as possible for long sequential content.Thus,RNNs can be used to compress text information in natural language processing.In this paper,a new neural network structure is designed with the advantage of RNNs,which can be used to determine the contextual correlation of text.The accuracy rate is over 80%.Compared with the previous work,the structure proposed in this paper guarantees the correctness and is more robust.In addition,this paper also designs a neural network structure that annotated polysyllabic words,which can be used to annotate polysyllabic words quickly and efficiently.The accuracy rate in polysyllabic words annotating reaches 99%.Finally,this paper proposes a lyrics generation system.After the production of different lyrics,the lyrics with high quality and corresponding poor quality lyrics are re-entered into the neural network so that they can continuously learn the contextual correlation of the lyrics.In this way,the system will be better in the context of contextual relevance of the text.