Attentive Statistics Pooling for Deep Speaker Embedding

Abstract

This paper proposes attentive statistics pooling for deep speaker embedding in text-independent speaker verification. In conventional speaker embedding, frame-level features are averaged over all the frames of a single utterance to form an utterance-level feature. Our method utilizes an attention mechanism to give different weights to different frames and generates not only weighted means but also weighted standard deviations. In this way, it can capture long-term variations in speaker characteristics more effectively. An evaluation on the NIST SRE 2012 and the VoxCeleb data sets shows that it reduces equal error rates (EERs) from the conventional method by 7.5% and 8.1%, respectively.

Proc. Interspeech 2018, pp2252--2256. arXiv admin note: text overlap with arXiv:1809.09311

References in corpus (3)

• VoxCeleb: a large-scale speaker identification dataset
• Deep Speaker: an End-to-End Neural Speaker Embedding System
• Feed-Forward Networks with Attention Can Solve Some Long-Term Memory Problems

Cited by in corpus (96)

• ECAPA-TDNN: Emphasized Channel Attention, Propagation and Aggregation in TDNN Based Speaker Verification
• In defence of metric learning for speaker recognition
• Clova Baseline System for the VoxCeleb Speaker Recognition Challenge 2020
• The IDLAB VoxSRC-20 Submission: Large Margin Fine-Tuning and Quality-Aware Score Calibration in DNN Based Speaker Verification
• Improving Multi-Scale Aggregation Using Feature Pyramid Module for Robust Speaker Verification of Variable-Duration Utterances
• Self-Supervised Learning Based Domain Adaptation for Robust Speaker Verification
• The SpeakIn System for VoxCeleb Speaker Recognition Challange 2021
• Spatial Pyramid Encoding with Convex Length Normalization for Text-Independent Speaker Verification
• Deep multi-metric learning for text-independent speaker verification
• SpeakerNet: 1D Depth-wise Separable Convolutional Network for Text-Independent Speaker Recognition and Verification
• Leveraging ASR Pretrained Conformers for Speaker Verification through Transfer Learning and Knowledge Distillation
• C3-DINO: Joint Contrastive and Non-contrastive Self-Supervised Learning for Speaker Verification
• Attention Back-end for Automatic Speaker Verification with Multiple Enrollment Utterances
• The VoxCeleb Speaker Recognition Challenge: A Retrospective
• Margin Matters: Towards More Discriminative Deep Neural Network Embeddings for Speaker Recognition
• The DKU-DukeECE Systems for VoxCeleb Speaker Recognition Challenge 2020
• NeXt-TDNN: Modernizing Multi-Scale Temporal Convolution Backbone for Speaker Verification
• A Unified Deep Learning Framework for Short-Duration Speaker Verification in Adverse Environments
• Combination of Deep Speaker Embeddings for Diarisation
• Channel adversarial training for speaker verification and diarization
• Utterance-level Aggregation For Speaker Recognition In The Wild
• The Accented English Speech Recognition Challenge 2020: Open Datasets, Tracks, Baselines, Results and Methods
• Meta-Learning for Short Utterance Speaker Recognition with Imbalance Length Pairs
• Lightweight Speaker Verification for Online Identification of New Speakers with Short Segments
• Exploring wav2vec 2.0 on speaker verification and language identification
• USAT: A Universal Speaker-Adaptive Text-to-Speech Approach
• Multi-target Extractor and Detector for Unknown-number Speaker Diarization
• RawNet: Advanced end-to-end deep neural network using raw waveforms for text-independent speaker verification
• An Adaptive X-vector Model for Text-independent Speaker Verification
• Speaker Recognition Based on Deep Learning: An Overview
• CN-Celeb: multi-genre speaker recognition
• Deep Normalization for Speaker Vectors
• SVSNet: An End-to-end Speaker Voice Similarity Assessment Model
• Toward Improving Synthetic Audio Spoofing Detection Robustness via Meta-Learning and Disentangled Training With Adversarial Examples
• Towards Lightweight Speaker Verification via Adaptive Neural Network Quantization
• Self-Adaptive Soft Voice Activity Detection using Deep Neural Networks for Robust Speaker Verification
• Improved RawNet with Feature Map Scaling for Text-independent Speaker Verification using Raw Waveforms
• Cosine Scoring with Uncertainty for Neural Speaker Embedding
• The 2020 Personalized Voice Trigger Challenge: Open Database, Evaluation Metrics and the Baseline Systems
• T-vectors: Weakly Supervised Speaker Identification Using Hierarchical Transformer Model
• Acoustic Scene Classification Using Fusion of Attentive Convolutional Neural Networks for DCASE2019 Challenge
• The NPU System for the 2020 Personalized Voice Trigger Challenge
• Beijing ZKJ-NPU Speaker Verification System for VoxCeleb Speaker Recognition Challenge 2021
• Adversarial Reweighting for Speaker Verification Fairness
• Bootstrap Equilibrium and Probabilistic Speaker Representation Learning for Self-supervised Speaker Verification
• Improving Noise Robustness In Speaker Identification Using A Two-Stage Attention Model
• Triplet Based Embedding Distance and Similarity Learning for Text-independent Speaker Verification
• Deep Speaker Vector Normalization with Maximum Gaussianality Training
• Optimizing Multi-Taper Features for Deep Speaker Verification
• End-to-end anti-spoofing with RawNet2
• I4U Submission to NIST SRE 2018: Leveraging from a Decade of Shared Experiences
• Domain-Invariant Speaker Vector Projection by Model-Agnostic Meta-Learning
• Voting for the right answer: Adversarial defense for speaker verification
• Spoofing-Aware Attention based ASV Back-end with Multiple Enrollment Utterances and a Sampling Strategy for the SASV Challenge 2022
• Deep Speaker Embedding Learning with Multi-Level Pooling for Text-Independent Speaker Verification
• Text-Independent Speaker Verification with Dual Attention Network
• Deep Discriminative Feature Learning for Accent Recognition
• Target Speaker Verification with Selective Auditory Attention for Single and Multi-talker Speech
• Self-supervised Text-independent Speaker Verification using Prototypical Momentum Contrastive Learning
• SEEF-ALDR: A Speaker Embedding Enhancement Framework via Adversarial Learning based Disentangled Representation
• Training Multi-Task Adversarial Network for Extracting Noise-Robust Speaker Embedding
• Supervised Speaker Embedding De-Mixing in Two-Speaker Environment
• An end-to-end approach for the verification problem: learning the right distance
• DropClass and DropAdapt: Dropping classes for deep speaker representation learning
• Phoneme-aware and Channel-wise Attentive Learning for Text DependentSpeaker Verification
• MACCIF-TDNN: Multi aspect aggregation of channel and context interdependence features in TDNN-based speaker verification
• Serialized Multi-Layer Multi-Head Attention for Neural Speaker Embedding
• Poformer: A simple pooling transformer for speaker verification
• End-to-End Language Identification using Multi-Head Self-Attention and 1D Convolutional Neural Networks
• Unsupervised Representation Learning for Speaker Recognition via Contrastive Equilibrium Learning
• Weakly Supervised Training of Hierarchical Attention Networks for Speaker Identification
• Neural Discriminant Analysis for Deep Speaker Embedding
• A Generalized Framework for Domain Adaptation of PLDA in Speaker Recognition
• Multi-Task Learning with High-Order Statistics for X-vector based Text-Independent Speaker Verification
• Speaker Verification By Partial AUC Optimization With Mahalanobis Distance Metric Learning
• Tongji University Undergraduate Team for the VoxCeleb Speaker Recognition Challenge2020
• Tongji University Team for the VoxCeleb Speaker Recognition Challenge 2020
• Multi-Task Network for Noise-Robust Keyword Spotting and Speaker Verification using CTC-based Soft VAD and Global Query Attention
• SoCov: Semi-Orthogonal Parametric Pooling of Covariance Matrix for Speaker Recognition
• Unsupervised Cross-Domain Speech-to-Speech Conversion with Time-Frequency Consistency
• Deep Neural Network Embeddings with Gating Mechanisms for Text-Independent Speaker Verification
• Enrollment-less training for personalized voice activity detection
• Speaker Representation Learning using Global Context Guided Channel and Time-Frequency Transformations
• Duality Temporal-channel-frequency Attention Enhanced Speaker Representation Learning
• An Improved Deep Neural Network for Modeling Speaker Characteristics at Different Temporal Scales
• Gaussian speaker embedding learning for text-independent speaker verification
• Parameterized Channel Normalization for Far-field Deep Speaker Verification
• Xi-Vector Embedding for Speaker Recognition
• MCSAE: Masked Cross Self-Attentive Encoding for Speaker Embedding
• Squeezing value of cross-domain labels: a decoupled scoring approach for speaker verification
• Supervised attention for speaker recognition
• CUHK-EE Systems for the vTAD Challenge at NCMMSC 2025
• STC speaker recognition systems for the NIST SRE 2021
• Inclusive Speaker Verification with Adaptive thresholding
• Learning Metrics from Mean Teacher: A Supervised Learning Method for Improving the Generalization of Speaker Verification System
• Improving robustness of one-shot voice conversion with deep discriminative speaker encoder