Proceedings Article10.1109/ICCE.2001.935284
Blind speech separation algorithm for dynamically mixing systems
Chi-Tat Leung,Wan-Chi Siu +1 more
- 19 Jun 2001
- pp 222-223
TL;DR: A variable rearrangement is derived to convert convolution operations into a simple matrix multiplication in dynamically mixing systems and the fast fixed-point algorithm is extended to separate a speech signal from background noise in a realistic room with acoustic reverberation.
read more
Abstract: This paper presents a blind speech separation algorithm that is capable of extracting a speech signal from background noise or music based on a microphone-array. A variable rearrangement is derived to convert convolution operations into a simple matrix multiplication in dynamically mixing systems. The fast fixed-point algorithm is then extended to separate a speech signal from background noise in a realistic room with acoustic reverberation.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
Speech Separation from Background of Music Based on Single-channel Recording
Xue-Cheng Jin,Zengfu Wang +1 more
- 20 Aug 2006
TL;DR: A speech separation algorithm, which is capable of extracting speech signals from music background when given only a single-channel recording, and can effectively remove background music from the input signals.
6
References
Fast and robust fixed-point algorithms for independent component analysis
TL;DR: Using maximum entropy approximations of differential entropy, a family of new contrast (objective) functions for ICA enable both the estimation of the whole decomposition by minimizing mutual information, and estimation of individual independent components as projection pursuit directions.
Convolutive blind separation of non-stationary sources
Lucas C. Parra,Clay D. Spence +1 more
TL;DR: This work tackles the problem of source separation by explicitly exploiting the nonstationarity of the acoustic sources, and finds an FIR backward model, which generates well separated model sources.
913
An experimental comparison of neural algorithms for independent component analysis and blind separation
TL;DR: Comparing five prominent neural or adaptive algorithms designed for Independent Component Analysis and blind source separation reveals robustness of the ICA algorithms with respect to modest modeling imperfections, and the superiority of fixed-point algorithms withrespect to the computational load.