Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Jul 15, 2026

Recording and Analyzing Multimodal Large-Scale Neuronal Ensemble Dynamics on CMOS-Integrated High-Density Microelectrode Array
09:44

Recording and Analyzing Multimodal Large-Scale Neuronal Ensemble Dynamics on CMOS-Integrated High-Density Microelectrode Array

Published on: March 8, 2024

Multi-scale spatio-temporal learning-based neural beamformer for multichannel speech enhancement.

Yanwen Li1, Huawei Chen1

  • 1College of Electronic and Information Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing, Jiangsu 211106, China.

The Journal of the Acoustical Society of America
|July 14, 2026
PubMed
Summary

This study introduces STBFNet, a novel neural beamforming network for multichannel speech enhancement. It effectively integrates spatial and temporal features for improved performance with reduced model size and latency.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same journal

Genetic algorithm-informed microjet control for noise reduction in supersonic impinging jetsa).

The Journal of the Acoustical Society of America·2026
Same journal

Physics-regularized neural acoustic fields for spatial layout inference from sparse room impulse responsesa).

The Journal of the Acoustical Society of America·2026
Same journal

Impedance eduction method based on multiple measurements with different incident modes in cylindrical ducts with reflective terminations.

The Journal of the Acoustical Society of America·2026
Same journal

Review of ultrasonic methods for monitoring, damage detection, and processing of lithium-ion batteries throughout their life cycle.

The Journal of the Acoustical Society of America·2026
Same journal

Assessment of silver nanoparticle mediated changes in osmotic fragility of red blood cells by a LASER diode based photoacoustic system.

The Journal of the Acoustical Society of America·2026
Same journal

Topographic effects on reflected acoustic waves from the OSIRIS-REx reentry observed from stratospheric balloons.

The Journal of the Acoustical Society of America·2026

Area of Science:

  • Signal Processing
  • Machine Learning
  • Speech Enhancement

Background:

  • Traditional time-domain neural beamformers use sequential networks, limiting joint spatio-temporal (ST) feature learning.
  • Existing methods struggle to capture rich ST representations crucial for multichannel speech enhancement.

Purpose of the Study:

  • To propose STBFNet, a novel time-domain neural beamforming network.
  • To enhance multichannel speech enhancement by effectively integrating multi-scale ST feature learning.

Main Methods:

  • Leveraging multi-scale convolutional operations for local ST feature extraction and aggregation.
  • Employing spatial information compensation to improve global channel interactions.
  • Utilizing self-attention and modified Conformer-style convolutions for refined ST representations.

Related Experiment Videos

Last Updated: Jul 15, 2026

Recording and Analyzing Multimodal Large-Scale Neuronal Ensemble Dynamics on CMOS-Integrated High-Density Microelectrode Array
09:44

Recording and Analyzing Multimodal Large-Scale Neuronal Ensemble Dynamics on CMOS-Integrated High-Density Microelectrode Array

Published on: March 8, 2024

Main Results:

  • STBFNet demonstrates superior performance compared to existing multichannel filtering methods.
  • The proposed network achieves a smaller model size and lower latency.
  • Competitive performance is maintained with enhanced ST representations.

Conclusions:

  • STBFNet offers an effective approach for time-domain neural beamforming.
  • The multi-scale ST feature learning strategy significantly improves speech enhancement.
  • The network provides an efficient solution for real-time applications.