Learning to evaluate: a few-shot non-intrusive speech quality assessment method based on meta-learning

Weili Zhou; Yuetao Liao; Tao Yang; Jinxiong Lai

doi:10.1117/12.3026310

8 April 2024 Learning to evaluate: a few-shot non-intrusive speech quality assessment method based on meta-learning

Weili Zhou, Yuetao Liao, Tao Yang, Jinxiong Lai

Proceedings Volume 13090, International Conference on Computer Application and Information Security (ICCAIS 2023); 130902A (2024) https://doi.org/10.1117/12.3026310
Event: International Conference on Computer Application and Information Security (ICCAIS 2023), 2023, Wuhan, China

Abstract

Speech Quality Assessment (SQA) is the process of evaluating and measuring the perceptual quality of speech signals. In recent years, the common challenge faced in deep learning is when there is insufficient labeled data for training effective models. Deep learning algorithms, especially deep neural networks, often require large amounts of labeled training data to generalize well to new, unseen examples. In this paper, a non-intrusive method based on meta-learning (Meta-SQA) is proposed for the few-shot learning of SQA. The speech samples which are corrupted by different channel distortions and environmental noises are first converted to the spectrum features then the training and testing tasks are formed for Meta- SQA. The U-Net with accurate localization of microscopic speech of neuronal structures is utilized as a meta-learner. The Meta-SQA obtains prior knowledge with the multiple SQA tasks of known channel distortions and achieves good adaption to the new channel distortions and environmental noises. In terms of person correlation and standard deviation of error measurements, this work achieves better performance than the compared methods.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Weili Zhou, Yuetao Liao, Tao Yang, and Jinxiong Lai "Learning to evaluate: a few-shot non-intrusive speech quality assessment method based on meta-learning", Proc. SPIE 13090, International Conference on Computer Application and Information Security (ICCAIS 2023), 130902A (8 April 2024); https://doi.org/10.1117/12.3026310

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
6 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Machine learning

Deep learning

Education and training

Data modeling

Molybdenum

Signal processing

Neural networks

Show All Keywords

Keywords/Phrases

Search In:

Publication Years