google/svq
Simple Voice Questions Simple Voice Questions (SVQ) is a set of short audio questions recorded in 26 locales across 17 languages under multiple audio conditions. It serves as a core evaluation componenet for Massive Sound Embedding Benchmark (MSEB). Technical Specifications Feature Details Locales 26 Languages 17 Total Speakers ~700 (Capped at 250 recordings per speaker) Audio Conditions Clean, Background Speech, Media, Traffic Noise Gender… See the full description on the dataset page: https://huggingface.co/datasets/google/svq.
5868k
document_retrieval_cross_lang.parquetdownload
document_retrieval_in_lang.parquetdownload
passage_retrieval_cross_lang.parquetdownload
passage_retrieval_in_lang.parquetdownload
query_reranking.parquetdownload
salient_term.parquetdownload
span_reasoning_cross_lang.parquetdownload
span_reasoning_in_lang.parquetdownload
span_retrieval_cross_lang.parquetdownload
span_retrieval_in_lang.parquetdownload
speech_transcription.parquetdownload
