Zobrazeno 1 - 10
of 799
pro vyhledávání: '"Oliveira, Danilo"'
Diffusion models have found great success in generating high quality, natural samples of speech, but their potential for density estimation for speech has so far remained largely unexplored. In this work, we leverage an unconditional diffusion model
Externí odkaz:
http://arxiv.org/abs/2410.17834
Generative speech enhancement has recently shown promising advancements in improving speech quality in noisy environments. Multiple diffusion-based frameworks exist, each employing distinct training objectives and learning techniques. This paper aims
Externí odkaz:
http://arxiv.org/abs/2409.10753
To obtain improved speech enhancement models, researchers often focus on increasing performance according to specific instrumental metrics. However, when the same metric is used in a loss function to optimize models, it may be detrimental to aspects
Externí odkaz:
http://arxiv.org/abs/2406.03460
Autor:
de Oliveira, Danilo, Gerkmann, Timo
Much research effort is being applied to the task of compressing the knowledge of self-supervised models, which are powerful, yet large and memory consuming. In this work, we show that the original method of knowledge distillation (and its more recen
Externí odkaz:
http://arxiv.org/abs/2309.09920
Autor:
Oliveira, Danilo Rogério de
A importância do Cerrado é devido a sua diversidade biológica e alto grau de endemismo, e demanda uma maior compreensão das formas de interação entre seus componentes. Atualmente, encontramos parte de sua estrutura original modificada pelo inte
Autor:
Oliveira, Danilo Carvalho
Objetivo: Analisar a Implantação do Componente Hospitalar da Rede de Urgências e Emergências- RUE na Região de Ribeirão Preto do Estado de São Paulo e propor uma matriz avaliativa. Método: Trata-se de um estudo avaliativo de análise de impla
Since its inception, the field of deep speech enhancement has been dominated by predictive (discriminative) approaches, such as spectral mapping or masking. Recently, however, novel generative approaches have been applied to speech enhancement, attai
Externí odkaz:
http://arxiv.org/abs/2306.03014
Esta dissertação aborda as transformações que a tecnologia digital está operando sobre aquilo que um dia chamaram corpo. Nossa atual capacidade de perceber, interagir e intervir em múltiplos espaços sem nos deslocarmos fisicamente revela que j
Publikováno v:
Proc. Interspeech 2023
In large part due to their implicit semantic modeling, self-supervised learning (SSL) methods have significantly increased the performance of valence recognition in speech emotion recognition (SER) systems. Yet, their large size may often hinder prac
Externí odkaz:
http://arxiv.org/abs/2305.19184