# Interspeech 2026: Audio-Based Understanding of Audiobook Narration Appeal

Supplementary materials for the paper ["Audio-Based Understanding of Audiobook Narration Appeal"](https://arxiv.org/abs/2607.02473), accepted to Interspeech 2026.

**Authors:** Shahar Elisha, Mariano Beguerisse-Díaz, Emmanouil Benetos

## Contents

This repository contains the following files, all referenced in the paper:

- **`supplementary_materials.pdf`** — Supplementary figures and tables referenced in the paper.
- **`dataset.csv`** — The dataset of audiobooks used in the study. The CSV contains metadata derived from [LibriVox](https://librivox.org/) (public domain) and the [Internet Archive](https://archive.org/).
- **`aggregate_egemaps.py`** — eGeMAPS low-level descriptors to functional set aggregation across audio chunks.