Application of ASV for Voice Identification after VC and Duration Predictor Improvement in TTS Models

TL;DR

This paper explores an automatic speaker verification system that extracts voice features for use in a multi-voice TTS pipeline, demonstrating effectiveness in verifying converted voices with an EER of 20.669 in SSTC challenge.

Contribution

It introduces a novel speaker verification approach focusing on embedding extraction of pitch, energy, and phoneme duration for improved voice authentication in TTS systems.

Findings

Achieved an EER of 20.669 in SSTC challenge for voice conversion verification.

Demonstrated the potential of embedding-based features in voice biometric security.

Enhanced verification accuracy for manipulated voice data.

Abstract

One of the most crucial components in the field of biometric security is the automatic speaker verification system, which is based on the speaker's voice. It is possible to utilise ASVs in isolation or in conjunction with other AI models. In the contemporary era, the quality and quantity of neural networks are increasing exponentially. Concurrently, there is a growing number of systems that aim to manipulate data through the use of voice conversion and text-to-speech models. The field of voice biometrics forgery is aided by a number of challenges, including SSTC, ASVSpoof, and SingFake. This paper presents a system for automatic speaker verification. The primary objective of our model is the extraction of embeddings from the target speaker's audio in order to obtain information about important characteristics of his voice, such as pitch, energy, and the duration of phonemes. This information is used in our multivoice TTS pipeline, which is currently under development. However, this model was employed within the SSTC challenge to verify users whose voice had undergone voice conversion, where it demonstrated an EER of 20.669.