Neural Acoustic-Phonetic Approach for Speaker Verification With Phonetic Attention Mask

Abstract

Traditional acoustic-phonetic approach makes use of both spectral and phonetic information when comparing the voice of speakers. While phonetic units are not equally informative, the phonetic context of speech plays an important role in speaker verification (SV). In this paper, we propose a neural acousticphonetic approach that learns to dynamically assign differentiated weights to spectral features for SV. Such differentiated weights form a phonetic attention mask (PAM). The neural acoustic-phonetic framework consists of two training pipelines, one for SV and another for speech recognition. Through the PAM, we leverage the phonetic information for SV. We evaluate the proposed neural acoustic-phonetic framework on the RSR2015 database Part III corpus, that consists of random digit strings. We show that the proposed framework with PAM consistently outperforms baseline with an equal error rate reduction of 13.45% and 10.20% for female and male data, respectively.

MFA: TDNN with Multi-scale Frequency-channel Attention for Text-independent Speaker Verification with Short Utterances

“…Speaker verification (SV) aims to verify the claimed identity of an enrolled speaker, in either a text-dependent or a textindependent manner [1]. The former is based on fixed short phrases [2], while the latter doesn't impose any restrictions on the phonetic content [3]. As the text is fixed in text-dependent SV, only a small amount of data is often used during training [4].…”