- OS: Ubuntu 17.10
- Language: Python 3.6
- Package: tensorflow 1.9
as its name
rnn method based on char
- using raw data (provided by our teacher)
- proformance looks stupid
- easily trapped in endless loop lyrics
- not coherent, not logical
- nearly no rhyme
- using data split (in ./data_split_bytone/)
- proformance looks stupid as well
- easily trapped in endless loop lyrics
- looks logical in someway
- rhyme the same with training data
my exploring based on file "df_all_pinyin_clear.csv" provided by data-cleaning group
splitting data into several files, each file contains lyrics with the same tone
hope you find them useful
not completed yet
paper assigned to me and YiWeiwen to read and give a talk
my note when reading the paper above
in English, without any picture