Deep Music & Lyric Visualizer

This visualizer uses BigGAN (Brock et al., 2018), Sentiment Analysis, and Semantic Similarity to visualize music and its lyrics. Recognition to the repo: https://github.com/msieg/deep-music-visualizer

Summary

The code I added decides which images from the ImageNet dataset to include in the video by finding either the Sentiment or Semantic similarity between the image labels and the lyrics.

Testing

1. Follow the guide from the this repo: https://github.com/msieg/deep-music-visualizer.
2. Add an mp3 song to the root folder (e.g. beethoven.mp3), as well as a text file (e.g. lyrics.txt) of the lyrics of the song.
3. Run the following: python visualize.py --song beethoven.mp3 --class_picker 3 --lyrics_file lyrics.txt
4. class_picker decides which type of method to use for picking the classes for the video. (1=Previous Classes; 2=Sentiment Analysis; 3=Contextual Similarity; 4=Sentiment + Contextual; 5=From Pickle File; Other=Random)