Matches in Ghent University Academic Bibliography for { <https://biblio.ugent.be/publication/01H1437ZEKN9F8247WZP2SDXQR> ?p ?o. }
Showing items 1 to 22 of
22
with 100 items per page.
- 01H1437ZEKN9F8247WZP2SDXQR classification C1.
- 01H1437ZEKN9F8247WZP2SDXQR date "2023".
- 01H1437ZEKN9F8247WZP2SDXQR language "eng".
- 01H1437ZEKN9F8247WZP2SDXQR type conference.
- 01H1437ZEKN9F8247WZP2SDXQR hasPart 01H1439HYMY8432F2KG279P04X.pdf.
- 01H1437ZEKN9F8247WZP2SDXQR hasPart 01H1439RFQSSZHHE1KRBW2752T.pdf.
- 01H1437ZEKN9F8247WZP2SDXQR subject "Technology and Engineering".
- 01H1437ZEKN9F8247WZP2SDXQR doi "10.1109/icassp49357.2023.10096868".
- 01H1437ZEKN9F8247WZP2SDXQR isbn "9781728163277".
- 01H1437ZEKN9F8247WZP2SDXQR presentedAt urn:uuid:c6020488-325d-46da-b5c1-74f0e6a63f31.
- 01H1437ZEKN9F8247WZP2SDXQR abstract "Weak harmonics of voiced speech segments are often lost during the process of noise suppression – especially at low SNRs. This leads to a distortion in the harmonic structure, and an accompanying loss in quality. In this paper, inspired by previous work on speech harmonic enhancement using statistical methods, we present a loss function component we term cepstral excitation manipulation (CEM) loss, which is constructed based on the fundamental frequency-related cepstral coefficients. This component can be introduced to the training of state-of-the-art architectures and its benefit is benchmarked, here, on CRUSE. Experiments show that the proposed loss function component nicely supplements standard loss functions and the harmonic structure is better preserved. On average, the best system improves by 0.4 on PESQ and 0.47 on DNSMOS compared to the noisy input. Substantial improvements are primarily in low SNRs (-5 dB to 5 dB) – the range for which harmonic recovery is most required.".
- 01H1437ZEKN9F8247WZP2SDXQR author 4118DE52-A6F4-11E7-8878-09B7AD28A064.
- 01H1437ZEKN9F8247WZP2SDXQR author 95CB9188-A74C-11E9-951A-2F1C5707D3EF.
- 01H1437ZEKN9F8247WZP2SDXQR dateCreated "2023-05-23T10:42:26Z".
- 01H1437ZEKN9F8247WZP2SDXQR dateModified "2024-07-09T14:40:32Z".
- 01H1437ZEKN9F8247WZP2SDXQR name "Aiding speech harmonic recovery in DNN-based single channel noise reduction using cepstral excitation manipulation (CEM) components".
- 01H1437ZEKN9F8247WZP2SDXQR pagination urn:uuid:cbc6f8f1-4888-4366-9c89-6f9c6d2b19ac.
- 01H1437ZEKN9F8247WZP2SDXQR publisher urn:uuid:53b67ce5-b200-467f-94c2-1cf641e089d1.
- 01H1437ZEKN9F8247WZP2SDXQR sameAs LU-01H1437ZEKN9F8247WZP2SDXQR.
- 01H1437ZEKN9F8247WZP2SDXQR sourceOrganization urn:uuid:da840044-e3bc-401e-afbc-985e1760a512.
- 01H1437ZEKN9F8247WZP2SDXQR sourceOrganization urn:uuid:e0a73fd7-02a4-4c15-8d80-72da4a8ef41f.
- 01H1437ZEKN9F8247WZP2SDXQR type C1.