| dc.contributor | 應數系 | - |
| dc.creator (作者) | 邱普照; 郭岳承 | - |
| dc.creator (作者) | Kow, Pu-Zhao;Kao, An-Hsien;Kuo, Yueh-Cheng | - |
| dc.date (日期) | 2025-12 | - |
| dc.date.accessioned | 9-Jan-2026 10:09:19 (UTC+8) | - |
| dc.date.available | 9-Jan-2026 10:09:19 (UTC+8) | - |
| dc.date.issued (上傳時間) | 9-Jan-2026 10:09:19 (UTC+8) | - |
| dc.identifier.uri (URI) | https://nccur.lib.nccu.edu.tw/handle/140.119/161018 | - |
| dc.description.abstract (摘要) | We evaluate two classical digital signal processing (DSP) techniques on the Helsinki Speech Challenge dataset and assess the efficiency of our approach by comparing its performance with that of the winning entries. Both our findings and those of the DTU team demonstrate that signal reconstruction can be achieved without resorting to artificial neural networks or deep learning. For Task 1, the results presented in this paper surpass those of the top three winners. Taken together, our results and those of the DTU team highlight the potential of neural-network-free approaches for speech signal reconstruction, offering lightweight yet effective alternatives to the deep learning-based methods commonly used in speech enhancement tasks. | - |
| dc.format.extent | 100 bytes | - |
| dc.format.mimetype | text/html | - |
| dc.relation (關聯) | Applied Mathematics for Modern Challenges, Vol.6, pp.15-23 | - |
| dc.subject (關鍵詞) | Pattern recognition; speech recognition; inverse problem; Fourier transform | - |
| dc.title (題名) | A neural-network-free approach to signal reconstruction | - |
| dc.type (資料類型) | article | - |
| dc.identifier.doi (DOI) | 10.3934/ammc.2025014 | - |
| dc.doi.uri (DOI) | https://doi.org/10.3934/ammc.2025014 | - |
