Biennial on Digital Signal Processing for In-Vehicle and Mobile Systems Sesimbra, Portugal September 2-3, 2005

Rapid Filter Adaptation for Frequency-Domain Independent Component Analysis in Various Car Noise Environments

Atsunobu Kaminuma (1), Daisuke Saitoh (1), Hiroshi Saruwatari (2), Tsuyoki Nishikawa (1), Akinobu Lee (2)

(1) Nissan Research Center, Natsushima-cho, Yokosuka-shi, Kanagawa, Japan
(2) Nara Institute of Science and Technology, Takayama-cho, Ikoma-shi, Nara, Japan

A computational complexity reduction method in a noise reduction algorithm using ICA in car components is described. We examined a noise suppression system and a speech enhancement system that use frequency-domain independent component analysis (hereafter FDICA) in car compartments with speech input systems. To achieve real-time processing in a car, we must reduce the computational complexity. We solved this problem with real-time processing by controlling the adaptive timing.

Bibliographic reference.  Kaminuma, Atsunobu / Saitoh, Daisuke / Saruwatari, Hiroshi / Nishikawa, Tsuyoki / Lee, Akinobu (2005): "Rapid filter adaptation for frequency-domain independent component analysis in various car noise environments", In DSP-in-V-2005, paper M2-5 (abstract).