ACA-Net: Towards Lightweight Speaker Verification using Asymmetric Cross Attention
Not on the map. Why?
Abstract (arXiv)
In this paper, we propose ACA-Net, a lightweight, global context-aware speaker embedding extractor for Speaker Verification (SV) that improves upon existing work by using Asymmetric Cross Attention (ACA) to replace temporal pooling. ACA is able to distill large, variable-length sequences into small, fixed-sized latents by attending a small query to large key and value matrices. In ACA-Net, we build a Multi-Layer Aggregation (MLA) block using ACA to generate fixed-sized identity vectors from variable-length inputs. Through global attention, ACA-Net acts as an efficient global feature extractor that adapts to temporal variability unlike existing SV models that apply a fixed function for pooling over the temporal dimension which may obscure information about the signal's non-stationary temporal variability. Our experiments on the WSJ0-1talker show ACA-Net outperforms a strong baseline by 5% relative improvement in EER using only 1/5 of the parameters.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 27 of its 36 references that are in the map. First posted on arXiv 20 May 2023.
Builds on
27 of 36 references.
References · 27
- 2017Attention Is All You Needcited by 60,361
- 2014Adam: A Method for Stochastic Optimizationcited by 37,240
- 2017Squeeze-and-Excitation Networkscited by 4,789
- 2017Deep Setscited by 1,421
- 2021Perceiver: General Perception with Iterative Attentioncited by 1,104
- 2015Cyclical Learning Rates for Training Neural Networkscited by 629
- 2021Perceiver IO: A General Architecture for Structured Inputs & Outputscited by 629
- 2020ECAPA-TDNN: Emphasized Channel Attention, Propagation and Aggregation in TDNN Based Speaker Verificationcited by 628
- 2019Res2Net: A New Multi-scale Backbone Architecturecited by 470
- 2021SpeechBrain: A General-Purpose Speech Toolkitcited by 406
- 2015Deep clustering: Discriminative embeddings for segmentation and separationcited by 212
- 2020Attention is All You Need in Speech Separationcited by 185
15 more
- 2018Attentive Statistics Pooling for Deep Speaker Embeddingcited by 114
- 2022General-purpose, long-context autoregressive modeling with Perceiver ARcited by 60
- 2022Pushing the limits of raw waveform speaker recognitioncited by 51
- 2021ECAPA-TDNN Embeddings for Speaker Diarizationcited by 47
- 2022MFA-Conformer: Multi-scale Feature Aggregation Conformer for Automatic Speaker Verificationcited by 47
- 2020SpEx: Multi-Scale Time Domain Speaker Extraction Networkcited by 30
- 2022Leveraging Modality-specific Representations for Audio-visual Speech Recognition via Reinforcement Learningcited by 22
- 2020Principled approach to the selection of the embedding dimension of networkscited by 19
- 2022Asymmetric Cross-Scale Alignment for Text-Based Person Searchcited by 18
- 2022Efficient Transformer-based Speech Enhancement Using Long Frames and STFT Magnitudescited by 10
- 2023deHuBERT: Disentangling Noise in a Self-supervised Model for Robust Speech Recognitioncited by 7
- 2021Target Speaker Verification with Selective Auditory Attention for Single and Multi-talker Speechcited by 5
- 2023Contrastive Speech Mixup for Low-resource Keyword Spottingcited by 4
- 2021Study on the temporal pooling used in deep neural networks for speaker verificationcited by 3
- 2022Magnitude-aware Probabilistic Speaker Embeddingscited by 3
Built on by
Citing papers per year: 2026: 1.
1 paper.