Speakers' voices are to a high degree individual. In the present paper we report about an ongoing research project in which we study how temporal characteristics of human speech (e.g. segmental or prosodic timing patterns, speech rhythmic characteristics and durational patterns of voicing) contribute to speaker individuality. We report about the creation of the TEVOID-Corpus (Temporal Voice Idiosyncrasy) that we are currently creating in our lab at Zurich University. 8 speakers producing 16 spontaneous sentences each are currently in the database which is rapidly growing. The paper gives an overview of the general ideas for the data collection and first results showing that there are significant rhythmic differences (%V, %VO, VarcoPeak) in spontaneously produced sentences between speakers of Zurich German.

Index Terms: speech rhythm, speaker idiosyncratic features, speaker identification, forensic phonetics