Show HN: KuralHub 1.0, Speech emotion datasets for 70 languages in one place

Category: ai-ml

Tags: speech-emotion-recognition, multilingual, dataset-catalog, benchmark, self-supervised-learning

Score: 6.0/10 (Innovation: 5, Technical: 5, Documentation: 8, Utility: 6)

KuralHub is a curated catalog of speech emotion recognition datasets across 70 languages, paired with a benchmark of 11 self-supervised speech models on 29 languages. It provides a reproducible training pipeline, per-run results, and a detailed paper (Interspeech 2026) highlighting cross-lingual performance gaps, making it a useful reference for multilingual SER research.

Target audience: speech and NLP researchers, ML engineers, data scientists

Repository: https://aaivu.github.io/KuralHub/datasets.html · Python · MIT · 4 stars

View on Hacker News