SeniorTalk

World's first Chinese conversational speech dataset for the very elderly (aged 75-85). Contains 55.53 hours of high-quality speech from 101 natural conversations involving 202 participants across 16 Chinese provinces, balanced for gender, region, and age. Enables ASR, speaker verification, speaker diarization, and speech editing research for under-represented elderly populations. Joint work with Nankai University. CC BY-NC-SA 4.0 license.