Code and results for: Does Acoustic Information Add Value Beyond Text?
收藏官方服务：
资源简介：
Supplementary code and results for the article "Does Acoustic Information Add Value Beyond Text? A Reproducible Evaluation of Late-Fusion Sentiment Analysis for Automated Call Analytics". The study compares a text-only model (BERT), an audio-only model (a 1D CNN on a frame-averaged MFCC sequence) and weighted late fusion for three-class sentiment classification on the MELD benchmark (official splits; 2,610 test utterances), with three random seeds and fusion-weight selection on the development set only.
提供机构：
Zenodo创建时间：
2026-09-25
