Results 61 to 70 of about 3,717 (149)
Self-supervised learning (SSL) models have achieved remarkable success in speaker verification tasks, yet their robustness to real-world audio degradation remains insufficiently characterized.
Ajan Ahmed, Masudul H. Imtiaz
doaj +1 more source
Abstract The term “feeding difficulties” (FD) encompasses a range of phenotypes characterized by inadequate food intake and/or inappropriate eating habits for a given age. Eosinophilic esophagitis (EoE) is a chronic, immune‐mediated condition often affecting children. It leads to esophageal dysmotility, potentially impacting feeding/eating.
Maria Ruano‐Zaragoza +14 more
wiley +1 more source
The quality (PESQ) and intelligibility (STOI) results in unseen noises.
The quality (PESQ) and intelligibility (STOI) results in unseen noises.
Amil Daraz (9667034) +3 more
core +1 more source
Rehaussement du signal de parole voisé par filtrage adaptatif des modes intrinsèques empiriques [PDF]
In this paper a new method for voiced speech enhancement combining the Empirical Mode Decomposition (EMD) and the Adaptive Center Weighted Average (ACWA) filter is introduced.
TURKI, Monia +5 more
core +1 more source
Tuning in to Kids Together: Piloting an Emotion‐Focused Coparenting Program
ABSTRACT Group parenting programs, including emotion‐focused programs, are effective at improving children's emotional and behavioral adjustment; however, the impact of these programs may be limited due to parents, typically mothers, attending sessions alone.
Christina C. Ambrosi +3 more
wiley +1 more source
In this paper, a multi-task learning U-shaped neural network (MTU-Net) is proposed and applied to single-channel speech enhancement (SE). The proposed MTU-based SE method estimates an ideal binary mask (IBM) or an ideal ratio mask (IRM) by extending the ...
Geon Woo Lee, Hong Kook Kim
doaj +1 more source
A neural coding method based on feature sensing
It has been shown that human neural coding is characterized by non‐linearity and that different neurons are more responsive to stimuli with specific features. Based on this, we propose a neural coding method based on feature sensing (feature sensing neural coding; FSNC). Taking speech signals as an example to simulate the process of auditory formation,
Dongbin He, Aiqun Hu, Kaiwen Sheng
wiley +1 more source
PESQ scores in seen noise sources for IRM training-target.
PESQ scores in seen noise sources for IRM training-target.
Amil Daraz (9667034) +3 more
core +1 more source
This article proposes a distributed minimum variance distortionless response (MVDR) beamformer for speech enhancement, where the network‐wide steering vector used for generating MVDR is estimated in a distributed way. This method is iteratively performed, and in each iteration, the steering matrix and the beamformer coefficient are simultaneously ...
Xingyue Cui, Rui Wang
wiley +1 more source
This meta‐analysis evaluates the response to placebo across multiple outcome domains in eosinophilic oesophagitis trials. While > 40% of patients improve clinically on placebo (influenced by patient age and randomisation ratio), most patients do not improve histologically or endoscopically, highlighting the importance of objective disease assessment ...
Angelica Rivas +9 more
wiley +1 more source

