How utkarshlimbachiya-03/genre-classification-of-film-music-using-emotion-annotations-and-audio-features works
This repository documents a research workflow for classifying film-music genres using audio features aligned with timestamped valence–arousal annotations. Its principal path runs from FME-24 clips and annotations through timestamp-centered feature extraction, leakage-conscious evaluation, and baseline classifiers. The README also describes statistical feature analysis and reported results. No source excerpts or implementation files are available in the provided tree, so implementation paths and undocumented connections are omitted; edges reflect README statements only.
Components 9
Data alignment
- FME-24 data
- Timestamp alignment
Feature analysis
- Audio features
- ANOVA tests
Classification
- Grouped splits
- Logistic regression
- RBF SVM
- Classification results
Other components
- Researcher
Connections 9
- Researcher → FME-24 data: uses dataset
README.md - FME-24 data → Timestamp alignment: provides annotations
README.md - Timestamp alignment → Audio features: aligns features
README.md - Audio features → ANOVA tests: tests significance
README.md - Audio features → Grouped splits: supplies features
README.md - Grouped splits → Logistic regression: trains baseline
README.md - Grouped splits → RBF SVM: trains baseline
README.md - Logistic regression → Classification results: reports accuracy
README.md - RBF SVM → Classification results: reports accuracy
README.md