Paper ID | E-3-3.3 |
Paper Title |
A Study on Geometrically Constrained IVA with Auxiliary Function Approach and VCD for In-Car Communication |
Authors |
Kana Goto, Li Li, Riki Takahashi, Shoji Makino, Takeshi Yamada, University of Tsukuba, Japan |
Session |
E-3-3: Advanced Signal Processing and Machine Learning for Audio and Speech Applications |
Time | Thursday, 10 December, 17:30 - 19:30 |
Presentation Time: | Thursday, 10 December, 18:00 - 18:15 Check your Time Zone |
|
All times are in New Zealand Time (UTC +13) |
Topic |
Speech, Language, and Audio (SLA): Special Session: Advanced Signal Processing and Machine Learning for Audio and Speech Applications |
Abstract |
In this paper, we apply a geometrically constrained independent vector analysis (GCIVA) method to an in-car speech enhancement system, and experimentally confirm its effectiveness in realistic environments. Specifically, we employ GCIVA with the auxiliary function approach and vectorwise coordinate descent (GCAV-IVA) to enhance the target speech in in-car communication, where multiple co-occurring speeches are recorded with a triangle microphone array. GCAV-IVA is a recently proposed geometrically constrained blind source separation method, which has shown to be powerful in directional speech enhancement with a limited number of microphones. Moreover, it is noteworthy in fast convergence, low computational cost, and no step-size tunning requirement, which makes it suitable for practical application. This paper investigates GCAV-IVA using measured in-car IRs to simulate more realistic environments. Moreover, we apply GCAV-IVA in a data-adaptive manner. The experimental results revealed that GCAV-IVA significantly outperformed the conventional beamforming methods in terms of signal-to-distortion ratios. |