Papers › Reshape Dimensions Network for Speaker Recognition

Reshape Dimensions Network for Speaker Recognition

25 Jul 2024arXiv:2407.18223archive 2025-07-28

Ivan Yakovlev, Rostislav Makarov, Andrei Balykin, Pavel Malov, Anton Okhotnikov, Nikita Torgashov

In this paper, we present Reshape Dimensions Network (ReDimNet), a novel neural network architecture for extracting utterance-level speaker representations. Our approach leverages dimensionality reshaping of 2D feature maps to 1D signal representation and vice versa, enabling the joint usage of 1D and 2D blocks. We propose an original network topology that preserves the volume of channel-timestep-frequency outputs of 1D and 2D blocks, facilitating efficient residual feature maps aggregation. Moreover, ReDimNet is efficiently scalable, and we introduce a range of model sizes, varying from 1 to 15 M parameters and from 0.5 to 20 GMACs. Our experimental results demonstrate that ReDimNet achieves state-of-the-art performance in speaker recognition while reducing computational complexity and the number of model parameters.

PaperPDFConference PDFCode

Syntology This paper's arXiv id is not in Syntology's Atlas id list (549,449 ids, as this build read them; the read time is in the build record), so this page links to no Syntology page for it. The list is not the whole graph; Syntology's feed shows papers as they enter the graph.

Code

IDRnD/ReDimNet officialmentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to DFKI-Speech System for WildSpoof Challenge: A robust framework for SASV In-the-Wild (arXiv:2602.02286; that paper's own run record: 5 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 2 violated, 1 with no contract checked; 2 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method. report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Speaker Recognition

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Speaker Verification VoxCeleb ReDimNet-B6-SF2-LM-ASNorm (15.0M) EER 0.37 #2 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B5-SF2-LM-ASNorm (9.2M) EER 0.39 #4 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B6-SF2-LM (15.0M) EER 0.4 #5 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B5-SF2-LM (9.2M) EER 0.43 #6 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B4-LM-ASNorm (6.3M) EER 0.44 #7 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B3-LM-ASNorm (3.0M) EER 0.47 #8 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B3-LM (3.0M) EER 0.5 #9 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B4-LM (6.3M) EER 0.51 #10 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B2-SF2-LM-ASNorm (4.7M) EER 0.52 #11 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B2-SF2-LM (4.7M) EER 0.57 #12 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B1-LM-ASNorm (2.2M) EER 0.73 #14 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B1-LM (2.2M) EER 0.85 #16 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B0-LM-ASNorm (1.0M) EER 1.07 #18 of 21 Archive leaderboard report
Speaker Verification VoxCeleb ReDimNet-B0-LM (1.0M) EER 1.16 #20 of 21 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B6-SF2-LM-ASNorm (15.0M) EER 0.37 #1 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B5-SF2-LM-ASNorm (9.2M) EER 0.39 #2 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B6-SF2-LM (15.0M) EER 0.4 #3 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B5-SF2-LM (9.2M) EER 0.43 #4 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B4-LM-ASNorm (6.3M) EER 0.44 #5 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B3-LM-ASNorm (3.0M) EER 0.47 #6 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B3-LM (3.0M) EER 0.5 #7 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B4-LM (6.3M) EER 0.51 #8 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B2-SF2-LM-ASNorm (4.7M) EER 0.52 #9 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B2-SF2-LM (4.7M) EER 0.57 #10 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B1-LM-ASNorm (2.2M) EER 0.73 #11 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B1-LM (2.2M) EER 0.85 #12 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B0-LM-ASNorm (1.0M) EER 1.07 #14 of 16 Archive leaderboard report
Speaker Verification VoxCeleb1 ReDimNet-B0-LM (1.0M) EER 1.16 #15 of 16 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections