Solutions for two important problems for the deployment of noise-robust large vocabulary automatic speech recognizers using the missing data paradigm are presented. irst problem is the generation of missing data masks. We propose and evaluate a method based on vector quantization and harmonicity that successfully exploits the characteristics of speech while requiring only weak assumptions on th...

We produce new examples of harmonic maps, having as either source or target manifold the tangent bundle TM on a Riemannian manifold (M,g), equipped with a Riemannian g-natural metric G. In particular, we study the harmonicity of the canonical projection π : (TM,G)→ (M,g), and of the identity map (TM,G)→ (TM,gS) and conversely, gS being the Sasaki metric on TM. A corresponding study is made for ...

Infants learn to use auditory and visual information to organize the sensory world into identifiable objects with particular locations. Here we use a behavioural method to examine infants' use of harmonicity cues to auditory object perception in a multisensory context. Sounds emitted by different objects sum in the air and the auditory system must figure out which parts of the complex waveform ...

The amount of masking of sounds from one source (signals) by sounds from a competing source (maskers) heavily depends on the sound characteristics of the masker and the signal and on their relative spatial location. Numerous studies investigated the ability to detect a signal in a speech or a noise masker or the effect of spatial separation of signal and masker on the amount of masking, but the...

I discuss three fundamental questions underpinning the study of consonance: 1) What features cause a particular chord to be perceived as consonant? 2) How did humans evolve ability perceive these features? 3) Why attribute aesthetic valences (if they at all)? The first question has been addressed by several recent articles, including Friedman, Kowalewski, Vuvan, and Neill (2021), with common co...