Identifying Repetitive Songs using LZ Compression
summary
Builds a dataset by scraped lyrics from six sources and measures repetitiveness with a custom LZ77 compressor that reports compression ratios (very repetitive songs compress to ~0.2–0.35 of original). Uses pandas + plotly to analyze Israeli radio charts back to the 1970s, finds no clear long trend but a downward slope since ~2012 and a sharp outlier before 2000, identifies most repetitive artists (e.g., Neta Barzilai) and less repetitive ones (many rappers), and ships an interactive web interface with the visualizations (code open except the lyrics).
AI-written summary · about
chapters (8)
- 0:00Introduction
- 3:56Song repetitiveness definition
- 7:10Lossless compression
- 10:02Artist repetitiveness analysis
- 13:35Data processing challenges
- 15:21Theoretical extensions
- 17:51Repetitiveness visualization
- 22:30Summary and website
related talks
30mAre we singing sillier?
27mPythonic Metal
59mClosing Keynote: Paul Lamere
35mDiversity and Data: Cases in the Music Industry
40mBuilding A Folk Music Recommendation System with LLMs
25mDIY Data Visualization in JavaScript while Stanning BTS
31mAnalyzing Rap Lyrics Using Python
15mTalk Less, Graph More: NLP, Networks and Hamilton Musical