Browsing by Department "University of Crete"

Now showing 1 - 4 of 4
  • Thumbnail Image
    Item
    Acoustic beamforming in front of a reflective plane
    (2018-11-29) Stefanakis, Nikolaos; Delikaris-Manias, Symeon; Mouchtaris, Athanasios; Hellenic Mediterranean University; Dept Signal Process and Acoust; University of Crete
    In this paper, we consider the problem of beamforming with a planar microphone array placed in front of a wall of the room, so that the microphone array plane is perpendicular to that of the wall. While this situation is very likely to occur in a real life problem, the reflections introduced by the adjacent wall can be the cause of a serious mismatch between the actual acoustic paths and the traditionally employed free-field propagation model. We present an adaptation from the free-field to the so-called reflection-aware propagation model, that exploits an in-situ estimation of the complex and frequency-dependent wall reflectivity. Results presented in a real environment demonstrate that the proposed approach may bring significant improvements to the beamforming process compared to the free-field propagation model, as well as compared to other reflection-aware models that have been recently proposed.
  • Thumbnail Image
    Item
    Applications of Spatially Localized Active-Intensity Vectors for Sound-Field Visualization
    (AUDIO ENGINEERING SOC, 2019-11) McCormack, Leo; Delikaris-Manias, Symeon; Politis, Archontis; Pavlidi, Despoina; Farina, Angelo; Pinardi, Daniel; Pulkki, Ville; Dept Signal Process and Acoust; University of Crete; University of Parma; Communication Acoustics: Spatial Sound and Psychoacoustics
    The purpose of this article is to detail and evaluate three alternative approaches to sound-field visualization, which all employ the use of spatially localized active-intensity (SLAI) vectors. These SLAI vectors are of particular interest, as they allow direction-of-arrival (DoA) estimates to be extracted in multiple spatially localized sectors, such that a sound source present in one sector has reduced influence on the DoA estimate made in another sector. These DoA estimates may be used to visualize the sound-field by either: (I) directly depicting the estimates as icons, with their relative size dictated by the corresponding energy of each sector; (II) generating traditional activity maps via histogram analysis of the DoA estimates; or (III) by using the DoA estimates to reassign energy and subsequently sharpen traditional beamformer-based activity maps. Since the SLAI-based DoA estimates are continuous, these approaches are inherently computationally efficient, as they forego the need for dense scanning grids to attain high-resolution imaging. Simulation results also show that these SLAI-based alternatives outperform traditional active-intensity and beamformer-based approaches, for the majority of cases.
  • Thumbnail Image
    Item
    The RoboPol pipeline and control system
    (2014) King, O. G.; Blinov, D.; Ramaprakash, A. N.; Myserlis, I.; Angelakis, E.; Baloković, M.; Feiler, R.; Fuhrmann, L.; Hovatta, T.; Khodade, P.; Kougentakis, A.; Kylafis, N.; Kus, A.; Modi, D.; Paleologou, E.; Panopoulou, G.; Papadakis, I.; Papamastorakis, I.; Paterakis, G.; Pavlidou, V.; Pazderska, B.; Pazderski, E.; Pearson, T. J.; Rajarshi, C.; Readhead, A. C S; Reig, P.; Steiakaki, A.; Tassis, K.; Zensus, J. A.; California Institute of Technology; St. Petersburg State University; Inter-University Centre for Astronomy and Astrophysics India; Max Planck Institute for Radio Astronomy; Nicolaus Copernicus University in Toruń; Metsähovi Radio Observatory; Foundation for Research and Technology - Hellas; University of Crete
    We describe the data reduction pipeline and control system for the RoboPol project. The RoboPol project is monitoring the optical R-band magnitude and linear polarization of a large sample of active galactic nuclei that is dominated by blazars. The pipeline calibrates and reduces each exposure frame, producing a measurement of the magnitude and linear polarization of every source in the 13 arcmin × 13 arcmin field of view. The control system combines a dynamic scheduler, real-time data reduction, and telescope automation to allow high-efficiency unassisted observations.
  • Thumbnail Image
    Item
    Speaker-independent raw waveform model for glottal excitation
    (2018-09-02) Juvela, Lauri; Tsiaras, Vassilis; Bollepalli, Bajibabu; Airaksinen, Manu; Yamagishi, Junichi; Alku, Paavo; Dept Signal Process and Acoust; University of Crete; National Institute of Informatics
    Recent speech technology research has seen a growing interest in using WaveNets as statistical vocoders, i.e., generating speech waveforms from acoustic features. These models have been shown to improve the generated speech quality over classical vocoders in many tasks, such as text-to-speech synthesis and voice conversion. Furthermore, conditioning WaveNets with acoustic features allows sharing the waveform generator model across multiple speakers without additional speaker codes. However, multi-speaker WaveNet models require large amounts of training data and computation to cover the entire acoustic space. This paper proposes leveraging the source-filter model of speech production to more effectively train a speaker-independent waveform generator with limited resources. We present a multi-speaker ’GlotNet’ vocoder, which utilizes a WaveNet to generate glottal excitation waveforms, which are then used to excite the corresponding vocal tract filter to produce speech. Listening tests show that the proposed model performs favourably to a direct WaveNet vocoder trained with the same model architecture and data.