20.500.12556/DKUM-91490 Brezizgubno stiskanje avdio posnetkov z nevronskimi mrežami magistrsko delo Lossless audio compression with neural networks Magistrska naloga se začne s kratkim pregledom relevantnih arhitektur nevronskih mrež in obstoječih brezizgubnih metod stiskanja avdia. Nato je predstavljena nova metoda za brezizgubno stiskanje avdia, ki temelji na napovedovanju naslednjega avdio vzorca s pomočjo konvolucijske nevronske mreže. Mreža se za vsak vhodni avdio posnetek uči posebej. Sledijo optimizacija hiperparametrov in nastavitev algoritma ter primerjava predlagane metode z obstoječimi algoritmi. The thesis begins with a brief overview of relevant neural network architectures and existing lossless audio compression methods. It then introduces a novel method for lossless audio compression, based on predicting the next audio sample using a convolutional neural network. The network is trained individually for each input audio recording. The research continues with optimizing algorithm hyperparameters and settings, followed by a comparison of the proposed method with existing algorithms. stiskanje algoritem entropija Golomb-Riceovo kodiranje strojno učenje compression algorithm entropy Golomb-Rice coding machine learning true true false [L. Železnik] Slovenski jezik Angleški jezik Magistrsko delo/naloga 2025-01-07 17:12:59 2025-02-06 09:17:32 2025-02-22 03:10:46 0000-00-00 00:00:00 2025 0 Maribor 0 Maribor 1 spletni vir (1 datoteka PDF (X, 68 str.)) 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.627:004.8.021(043.2) 226893315 MAG_Zeleznik_Luka_2025.pdf MAG_Zeleznik_Luka_2025.pdf 1 BEE057F2913CBA2D2E558A8AAE4E6A2B 09b7132c1c534d3062f7364c99bf629fddf209d00d596812cfdbd9a185fb6fe8 566dd313-cd24-11ef-8aab-00155d000105 https://dk.um.si/Dokument.php?lang=slv&id=187671 MAG_Zeleznik_Luka_2025.zip MAG_Zeleznik_Luka_2025.zip 1 6889F2627B5C4DC46B4310B9F221A877 5778b6df0c15f684e5c7719230c761cffb8b27074ed7f0c96738230ea8da1198 6a0b50f0-cd24-11ef-8aab-00155d000105 https://dk.um.si/Dokument.php?lang=slv&id=187672 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0