Speech enables easy human-to-human communication as well as human-to-machine
interaction. However, the quality of speech degrades due to background noise in the …

The paper provides a comprehensive overview of speech enhancement techniques and
their applications. It discusses challenges in non-stationary noise, reverberation, and …

Abstract We propose the Multi-Head Gaussian Adaptive Attention Mechanism (GAAM), a
novel probabilistic attention framework, and the Gaussian Adaptive Transformer (GAT) …

In today's scientific epoch, speech is an important means of communication. Speech
enhancement is necessary for increasing the quality of speech. However, the presence of …

Extracting robust effective speech features is one of the challenging topics in the speaker
recognition field, especially in noisy conditions. It can substantially improve the robustness …

We propose the Multi-Head Density Adaptive Attention Mechanism (DAAM), a novel
probabilistic attention framework that can be used for Parameter-Efficient Fine-tuning …

This paper addresses the common challenge of system performance degradation due to
speech inconsistency and mismatched acoustic conditions across various domains in …

Speech is a signal produced by humans to interact and communicate. Different information
is gained from speech signals, such as the language being spoken, emotion, gender …

Speech enhancement is an essential and challenging task in the signal-processing field. In
this article, a multiframe maximum likelihood distortionless response (MFMLDR) filter is …

With the exponential increase of mobile users and internet subscribers, the utilization of
voice over internet protocol (VoIP) application is increasing dramatically. People exploit …