SpeakerID is a corpus of 100 spoken sentences and pseudosentences in European Portuguese (PT) and Mandarin Chinese (CH) designed to enable research on speaker identity.
The utterances were recorded by five male speakers of European Portuguese (Speakers A-E) and five male speakers of Mandarin Chinese (Speakers F-J). Each speaker recorded 10 sentences (Chinese) or 10 pseudosentences (Portuguese) resulting in a set of 50 stimuli per language. Edited soundfiles were normalized for amplitude at 65 dB. The two stimulus sets are similar in other basic acoustic parameters, namely duration, mean fundamental frequency, and standard deviation of mean fundamental frequency.
SpeakerID contains an appendix in Excel format presenting the main characteristics of the soundfiles. The database provides stimuli that are useful for research on voice processing, namely voice identification, and also for cross-cultural and linguistic comparisons.