Datasets:

---

MERaLiON

/

sea_audiobench_datasets_TCQ

Like 0

Follow

MERaLiON 131

Adds a dataset card for the TCQ (timestamped content query) task of SEA-SpeechBench.

Contents:

• YAML metadata: license, languages, task categories, source datasets, and gated-access notice for the non-commercial source material
• Loading example for datasets>=4.0, with a note on config size
• Subset table: 27 configs, 14,172 examples, source corpus, context length and audio duration per config
• Data field descriptions
• Source data table with upstream licenses
• BibTeX for the benchmark and the three source corpora