Exploiting spatial information with the informed complex-valued spatial autoencoder for target speaker extraction
Briegleb A, Halimeh MM, Kellermann W (2023)
Publication Language: English
Publication Type: Conference contribution, Conference Contribution
Publication year: 2023
Event location: Rhodes, Greece
URI: https://ieeexplore.ieee.org/document/10095196
DOI: 10.1109/ICASSP49357.2023.10095196
Open Access Link: https://arxiv.org/abs/2210.15512
Abstract
In conventional multichannel audio signal enhancement, spatial and spectral filtering are often performed sequentially. In contrast, it has been shown that for neural spatial filtering a joint approach of spectro-spatial filtering is more beneficial. In this contribution, we investigate the spatial filtering performed by such a time-varying spectro-spatial filter. We extend the recently proposed complex-valued spatial autoencoder (COSPA) for the task of target speaker extraction by leveraging its interpretable structure and purposefully informing the network of the target speaker’s position. We show that the resulting informed COSPA (iCOSPA) effectively and flexibly extracts a target speaker from a mixture of speakers. We also find that the proposed architecture is well capable of learning pronounced spatial selectivity patterns and show that the results depend significantly on the training target and the reference signal when computing various evaluation metrics.
Authors with CRIS profile
Annika Briegleb
Professur für Signalverarbeitung
Mhd Modar Halimeh
Professur für Signalverarbeitung
Walter Kellermann
Professur für Signalverarbeitung
How to cite
APA:
Briegleb, A., Halimeh, M.M., & Kellermann, W. (2023). Exploiting spatial information with the informed complex-valued spatial autoencoder for target speaker extraction. In IEEE (Eds.), Proceedings of the 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Rhodes, Greece, GR.
MLA:
Briegleb, Annika, Mhd Modar Halimeh, and Walter Kellermann. "Exploiting spatial information with the informed complex-valued spatial autoencoder for target speaker extraction." Proceedings of the 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Rhodes, Greece Ed. IEEE, 2023.
BibTeX: Download