This repository provides the audio part of the CommonVoice-SpeechRE dataset, a benchmark for Speech Relation Extraction (SpeechRE), presented in the paper CommonVoice-SpeechRE and RPG-MoGe: Advancing Speech Relation Extraction with a New Dataset and Multi-Order Generative Framework.
It contains 19,583 speech samples derived from Common Voice 17.0.
All audio files are downsampled to 16kHz for consistency with common speech processing pipelines.
👉 The… See the full description on the dataset page:
https://huggingface.co/datasets/DMU-ITREC/CommonVoice-SpeechRE-audio.
