add a user controllable midi velocity
the default value for midi velocity is 120, it is the best value for model, while in processing, with input midi file different, adjusted velocity will make a better extracted melody
In one word, higher the velocity value, the melody extracted will be faster when play.

improve accuracy for extracting melody from any music
use a new , better TensorFlow model to predict data
add a new localization - English
the text description on AI's process more concrete and accuracy

优化了模型的预测
对钢琴曲的MIDI转换比之前更加精确
更改了部分算法
对于一般音乐的转换MIDI仍有局限性
模型是基于大量钢琴曲进行训练的，虽然可以用于一般音乐预测，但是对于复杂混声的音乐预测转换效果明显不太理想
建议使用电脑库乐队播放转换的MIDI文件，播放效果比手机直接播放要好很多