Harness the linguistic depth of the Chinese-speaking world with our China Speech Dataset collection. Spanning Mandarin, Cantonese, Wu, Min, Hakka, and dozens of regional topolects — across mainland China, Taiwan, Hong Kong, and diaspora communities — these datasets are built for teams developing voice AI that reflects how over a billion people truly communicate.Each recording is sourced from native speakers across diverse acoustic environments — from bustling megacities and rural villages to broadcast studios and spontaneous conversational settings. Meticulously annotated with tonal markers, character-level transcriptions, pinyin alignments, and speaker demographics, our Chinese collections are engineered for the phonetic precision and tonal complexity that Mandarin and its sister languages demand.