Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2402.07619

Developing a Multi-variate Prediction Model For COVID-19 From Crowd-sourced Respiratory Voice Data

Yuyang Yan·Wafaa Aljbawi·Sami O. Simons·Visara Urovi
View original source ↗Read source document ↗Export metadata ↓

Abstract

COVID-19 has affected more than 223 countries worldwide and in the Post-COVID Era, there is a pressing need for non-invasive, low-cost, and highly scalable solutions to detect COVID-19. We develop a deep learning model to identify COVID-19 from voice recording data. The novelty of this work is in the development of deep learning models for COVID-19 identification from only voice recordings. We use the Cambridge COVID-19 Sound database which contains 893 speech samples, crowd-sourced from 4352 participants via a COVID-19 Sounds app. Voice features including Mel-spectrograms and Mel-frequency cepstral coefficients (MFCC) and CNN Encoder features are extracted. Based on the voice data, we develop deep learning classification models to detect COVID-19 cases. These models include Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) and Hidden-Unit BERT (HuBERT). We compare their predictive power to baseline machine learning models. HuBERT achieves the highest accuracy of 86\% and the highest AUC of 0.93. The results achieved with the proposed models suggest promising results in COVID-19 diagnosis from voice recordings when compared to the results obtained from the state-of-the-art.

Explore related subjects

cs.SDcs.AIeess.AS

Keep this discovery

BibTeXRIS

Yuyang Yan, Wafaa Aljbawi, Sami O. Simons, Visara Urovi. 2024-02-12. Developing a Multi-variate Prediction Model For COVID-19 From Crowd-sourced Respiratory Voice Data. https://doi.org/10.37349/edht.2024.00022

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Xiaomi-CocktailASR-1 Technical Report

Recently, large language model (LLM) based ASR models have achieved significant progress, yet they generally lack support for multi-speaker scenarios, where the cocktail party problem remains a critical bottleneck for further advancing ASR. Existing TS-ASR methods, including end-to-end architectures with speaker embeddings and latest LLM-based explorations suffer from degraded single-speaker performance and the inability to reject when the target speaker is absent. In this paper, we propose Xiaomi-CocktailASR-1, an LLM-based end-to-end TS-ASR architecture. By utilizing reference speech as voiceprint prompts, it directly transcribes the target speaker's speech without requiring speech separation. Xiaomi-CocktailASR-1 maintains competitive performance in single-speaker scenarios, comparable to mainstream ASR models. It also features a negative sample rejection capability, outputting empty text when the target speaker is absent from the mixed speech. Additionally, Xiaomi-CocktailASR-1 supports a Chain-of-Thought (CoT) reasoning mode to provide explicit reasoning steps. Extensive experiments on various synthetic and real-world multispeaker benchmarks demonstrate that Xiaomi-CocktailASR-1 achieves state-of-the-art performance, effectively addressing the cocktail party problem through a unified architecture that balances multispeaker and single-speaker recognition accuracy, along with rejection capability.

cs.SD↗
arXiv2026

EConv-TasNet: Efficient Conv-TasNet for Effective Speech Separation

Conv-TasNet has served as a strong baseline for time-domain speech separation, and many studies have extended it with advanced architectures such as dual-path networks, U-Nets, and attention mechanisms. However, these methods often introduce high computational cost and complexity, limiting their deployment in resource-constrained scenarios. To address this issue, we propose eConv-TasNet, an efficient variant of Conv-TasNet that improves both effectiveness and efficiency without relying on resource-intensive modules. The proposed model consists of a group-wise early-splitting (GES) module and a multi-group feature aggregation (MGFA) module. GES generates discriminative speaker embeddings at intermediate stages, while MGFA progressively aggregates these group-level representations for refined mask estimation. Experimental results show that eConv-TasNet reduces model size by 22.4%, accelerates inference by 18.9%, and improves SI-SNRi by 14.0%-28.0% across three public benchmarks. Moreover, it achieves competitive performance compared with state-of-the-art methods while requiring significantly fewer parameters and lower inference cost. These results demonstrate a favorable efficiency-effectiveness trade-off for edge deployment.

cs.SD↗