
This paper describes a winning approach to the TdSV Challenge 2024, demonstrating that competitive performance in text-dependent speaker verification can be achieved using independent pre-trained models for phrase and speaker verification, without relying on joint modeling of speaker and text.


coremsg

bilingual-text-dependent-speaker-verification-using-pre-trained-models-a-winning-approach-at-the-tdsv-challenge-2024


Bilingual Text-dependent Speaker Verification Using Pre-trained Models: A Winning Approach at the TdSV Challenge 2024


title_rewrite


The TdSV Challenge 2024 aims to motivate participants to develop competitive systems and explore innovative concepts for text-dependent speaker verification, focusing on two distinct scenarios: conventional TdSV and speaker enrollment using user-defined passphrases.


text-dependent-speaker-verification-tdsv-challenge-2024-evaluation-plan-and-tasks


Text-Dependent Speaker Verification (TdSV) Challenge 2024: Evaluation Plan and Tasks
