A harmonic excitation state-space approach to blind separation of speech

Rasmus Kongsgaard Olsson, Lars Kai Hansen

    Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

    56 Downloads (Pure)

    Abstract

    We discuss an identification framework for noisy speech mixtures. A block-based generative model is formulated that explicitly incorporates the time-varying harmonic plus noise (H+N) model for a number of latent sources observed through noisy convolutive mixtures. All parameters including the pitches of the source signals, the amplitudes and phases of the sources, the mixing filters and the noise statistics are estimated by maximum likelihood, using an EM-algorithm. Exact averaging over the hidden sources is obtained using the Kalman smoother. We show that pitch estimation and source separation can be performed simultaneously. The pitch estimates are compared to laryngograph (EGG) measurements. Artificial and real room mixtures are used to demonstrate the viability of the approach. Intelligible speech signals are re-synthesized from the estimated H+N models.
    Original languageEnglish
    Title of host publicationAdvances in Neural Information Processing Systems
    PublisherMIT Press
    Publication date2005
    Pages993-1000
    Publication statusPublished - 2005

    Cite this

    Olsson, R. K., & Hansen, L. K. (2005). A harmonic excitation state-space approach to blind separation of speech. In Advances in Neural Information Processing Systems (pp. 993-1000). MIT Press. http://www2.imm.dtu.dk/pubdb/p.php?3340