A Comparative Re-assessment Of Feature Extractors For Deep Speaker Embeddings

Abstract

Modern automatic speaker verification relies largely on deep neural networks (DNNs) trained on mel-frequency cepstral coefficient (MFCC) features. While there are alternative feature extraction methods based on phase, prosody and long-term temporal operations, they have not been extensively studied with DNN-based methods. We aim to fill this gap by providing extensive re-assessment of 14 feature extractors on VoxCeleb and SITW datasets. Our findings reveal that features equipped with techniques such as spectral centroids, group delay function, and integrated noise suppression provide promising alternatives to MFCCs for deep speaker embeddings extraction. Experimental results demonstrate up to 16.3% (VoxCeleb) and 25.1% (SITW) relative decrease in equal error rate (EER) to the baseline.

Stats

• citations11
• S2 citations—
• github stars0
• HF likes0
• heat score8.09
• arxiv keyliu2020a

Related papers

• How To Improve Your Speaker Embeddings Extractor In Generic Toolkits (2018)9.76
• Deep Speaker Embeddings For Far-field Speaker Recognition On Short Utterances (2020)11.29
• Margin Matters: Towards More Discriminative Deep Neural Network Embeddings For Speaker Recognition (2019)15.25
• Feature Enhancement With Deep Feature Losses For Speaker Verification (2019)10.61
• Deep Speaker Embedding Learning With Multi-level Pooling For Text-independent Speaker Verification (2019)0.00
• On Deep Speaker Embeddings For Text-independent Speaker Recognition (2018)11.93
• Disentangled Speaker And Nuisance Attribute Embedding For Robust Speaker Verification (2020)8.60
• The Exploitation Of Multiple Feature Extraction Techniques For Speaker Identification In Emotional States Under Disguised Voices (2021)2.26