论文信息 - The multi-session audio research project (MARP) corpus: goals, design and initial findings

The multi-session audio research project (MARP) corpus: goals, design and initial findings

Abstract : This project describes the composition and goals of the Multi-session Audio Research Project (MARP) corpus and some initial experimental findings. The MARP corpus is a three year longitudinal collection of 21 sessions and more than 60 participants. This study was undertaken to test the impact of various factors on speaker recognition, such as inter-session variability, intonation, aging, whispering and text dependency. Initial results demonstrate the impact of sentence intonation, whispering, text dependency and cross session tests. These results highlight the sensitivity of speaker recognition to vocal, environmental and phonetic conditions that are commonly encountered but rarely explored or tested.

[1] Douglas A. Reynolds,et al. Comparison of background normalization methods for text-independent speaker verification , 1997, EUROSPEECH.

[2] Ayako Ikeno. Perceptual in-set speaker identification using neutral speech and speech under Lombard effect , 2006 .

[3] Aaron D. Lawson,et al. Perturbation and pitch normalization as enhancements to speaker recognition , 2009, 2009 IEEE International Conference on Acoustics, Speech and Signal Processing.

[4] Aaron Lawson,et al. External factors impacting the performance of speaker identification: Multisession audio research project (MARP) corpus experiments , 2007 .

[5] James R. Glass,et al. A Comparative Study of Methods for Handheld Speaker Verification in Realistic Noisy Conditions , 2006, 2006 IEEE Odyssey - The Speaker and Language Recognition Workshop.

[6] Douglas A. Reynolds,et al. Channel robust speaker verification via feature mapping , 2003, 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03)..

[7] Aaron D. Lawson,et al. Triphone-based confidence system for speaker identification , 2004, INTERSPEECH.