RLAC Audio Segmenter - Chroniques
Description
This is version 0.1 of a Random Forest classifier designed for radio audio segmentation. It identifies specific audio segments (columns, features, or ads) within long radio broadcasts.
Model Details
- Type: Random Forest Classifier
- Input: 3-second audio segments
- Features: MFCC (13), Spectral Energy (4 bands), Zero-Crossing Rate, RMS, Spectral Centroid, Rolloff, and Bandwidth.
- Version: v0.1
Usage
The model is trained to distinguish between targeted content and background broadcast material. It uses a confidence threshold of 0.89 to minimize false positives during the detection phase.
Author
Maintained by eglantinefonrose.
