Deep Speaker Feature Learning For Text-independent Speaker Verification

Abstract

Recently deep neural networks (DNNs) have been used to learn speaker features. However, the quality of the learned features is not sufficiently good, so a complex back-end model, either neural or probabilistic, has to be used to address the residual uncertainty when applied to speaker verification, just as with raw features. This paper presents a convolutional time-delay deep neural network structure (CT-DNN) for speaker feature learning. Our experimental results on the Fisher database demonstrated that this CT-DNN can produce high-quality speaker features: even with a single feature (0.3 seconds including the context), the EER can be as low as 7.68%. This effectively confirmed that the speaker trait is largely a deterministic short-time property rather than a long-time distributional pattern, and therefore can be extracted from just dozens of frames.

Stats

• citations46
• S2 citations—
• github stars0
• HF likes0
• heat score12.54
• arxiv keyli2017deep

Related papers

• Full-info Training For Deep Speaker Feature Learning (2017)7.16
• Deep Speaker Embedding Learning With Multi-level Pooling For Text-independent Speaker Verification (2019)0.00
• Feature Enhancement With Deep Feature Losses For Speaker Verification (2019)10.61
• DNN Based Speaker Recognition On Short Utterances (2016)0.00
• FDN: Finite Difference Network With Hierarchical Convolutional Features For Text-independent Speaker Verification (2021)0.00
• Neural Network Based Speaker Classification And Verification Systems With Enhanced Features (2017)8.60
• Deep CNN Based Feature Extractor For Text-prompted Speaker Recognition (2018)7.81
• Speakernet: 1D Depth-wise Separable Convolutional Network For Text-independent Speaker Recognition And Verification (2020)0.00