View source for 20240022864. DEEP LEARNING-BASED METHOD FOR ACOUSTIC FEEDBACK SUPPRESSION IN CLOSED-LOOP SYSTEM simplified abstract (INSTITUTE OF ACOUSTICS, CHINESE ACADEMY OF SCIENCES)

=DEEP LEARNING-BASED METHOD FOR ACOUSTIC FEEDBACK SUPPRESSION IN CLOSED-LOOP SYSTEM=


==Organization Name==

[[:Category:INSTITUTE OF ACOUSTICS, CHINESE ACADEMY OF SCIENCES|INSTITUTE OF ACOUSTICS, CHINESE ACADEMY OF SCIENCES]]


[[Category:INSTITUTE OF ACOUSTICS, CHINESE ACADEMY OF SCIENCES]]

==Inventor(s)==

[[:Category:Chengshi Zheng of Beijing (CN)|Chengshi Zheng of Beijing (CN)]][[Category:Chengshi Zheng of Beijing (CN)]]

[[:Category:Xiaodong Li of Beijing (CN)|Xiaodong Li of Beijing (CN)]][[Category:Xiaodong Li of Beijing (CN)]]

==DEEP LEARNING-BASED METHOD FOR ACOUSTIC FEEDBACK SUPPRESSION IN CLOSED-LOOP SYSTEM - A simplified explanation of the abstract==


This abstract first appeared for US patent application 20240022864 titled 'DEEP LEARNING-BASED METHOD FOR ACOUSTIC FEEDBACK SUPPRESSION IN CLOSED-LOOP SYSTEM

==Simplified Explanation==

The abstract describes a deep learning-based method for suppressing acoustic feedback in a closed-loop system. The method involves using an offline trained suppression model to process an audio signal input and then feeding the processed audio signal to a sound reproduction unit for playback, thereby achieving acoustic feedback suppression. The closed-loop system suppression model is built based on deep learning techniques. Additionally, the method involves modeling the closed-loop system, simulating the acoustic feedback path, and calculating a maximum stable gain for each simulated unit impulse response. A closed-loop signal is generated based on the maximum stable gain, and an open-loop target signal is generated using the audio signal input. The model is trained using parallel training data consisting of the closed-loop signal and open-loop target signal.

* The method uses a deep learning-based suppression model to suppress acoustic feedback in a closed-loop system.
* The closed-loop system is modeled and the maximum stable gain is calculated for each simulated unit impulse response.
* An open-loop target signal is generated using the audio signal input.
* The model is trained using parallel training data consisting of the closed-loop signal and open-loop target signal.

Potential applications of this technology:
* Audio systems in live performances or public address systems where acoustic feedback can be a common issue.
* Teleconferencing or video conferencing systems to improve audio quality and reduce feedback during meetings.
* Home theater systems or audio equipment to enhance the listening experience and prevent feedback.

Problems solved by this technology:
* Acoustic feedback can be a significant problem in closed-loop audio systems, causing distortion and unwanted noise.
* Traditional methods of feedback suppression may not be effective or may require manual adjustments.
* This technology provides an automated and efficient solution for suppressing acoustic feedback in a closed-loop system.

Benefits of this technology:
* Improved audio quality and clarity by reducing or eliminating acoustic feedback.
* Enhanced user experience in various audio applications.
* Automated feedback suppression without the need for manual adjustments or interventions.


==Original Abstract Submitted==

a deep learning-based method for acoustic feedback suppression in a closed-loop system, the method includes applying an offline trained closed-loop system suppression model, processing an audio signal input, and then feeding the processed audio signal to a sound reproduction unit of the closed-loop system for playback to achieve acoustic feedback suppression, the closed-loop system suppression model being built based on deep learning; and modeling the closed-loop system, generating a unit impulse response of an acoustic feedback path by simulation, and calculating a maximum stable gain according to each simulated unit impulse response, and generating a closed-loop signal based on the maximum stable gain; generating an open-loop target signal under an open-loop condition by using the audio signal input to the closed-loop system; forming parallel training data of the model by putting the closed-loop signal and open-loop target signal together, and training the model by using the generated parallel training data.



[[Category:H04R25/00]]
[[Category:G06N3/08]]