We propose a new front-end that reflects some aspects of auditory nerve response. Namely, the pattern of synchrony responses observed over auditory nerve fibers associated with F0, F1 and F2 of voiced sounds. The main goal is to get a set of features, which represents those frequency trajectories. These features should be less sensitive to adverse environmental conditions than mel-cepstrum or Fourier based front-ends. The core of the system is a computational implementation of the well-known electronic device phase locked loop (PLL). Outputs of a PLL bank interpolated and conveniently resampled define our front-end coefficients. Performance of the system was assessed in a vowels recognition task using a common HMM back-end, and compared with the performance of a mel-frequency cepstral coefficients (MFCC) front-end.
Cite as: Estienne, C., Pelle, P. (2000) A synchrony front-end using phase-locked-loop techniques. Proc. 6th International Conference on Spoken Language Processing (ICSLP 2000), vol. 3, 98-101, doi: 10.21437/ICSLP.2000-487
@inproceedings{estienne00_icslp,
author={Claudio Estienne and Patricia Pelle},
title={{A synchrony front-end using phase-locked-loop techniques}},
year=2000,
booktitle={Proc. 6th International Conference on Spoken Language Processing (ICSLP 2000)},
pages={vol. 3, 98-101},
doi={10.21437/ICSLP.2000-487}
}