Vo-Ve: An Explainable Voice-Vector for Speaker Identity Evaluation

TL;DR

Vo-Ve introduces an explainable voice-vector embedding that captures speaker identity and provides interpretable voice attribute probabilities, improving speaker similarity evaluation and enhancing speech task analysis.

Contribution

The paper presents Vo-Ve, a novel explainable voice-vector embedding that incorporates explicit voice attribute probabilities for better interpretability and evaluation.

Findings

Vo-Ve performs competitively in speaker similarity evaluation.

Vo-Ve offers high-level interpretability through voice attribute probabilities.

The approach enhances speech task evaluation schemes.

Abstract

In this paper, we propose Vo-Ve, a novel voice-vector embedding that captures speaker identity. Unlike conventional speaker embeddings, Vo-Ve is explainable, as it contains the probabilities of explicit voice attribute classes. Through extensive analysis, we demonstrate that Vo-Ve not only evaluates speaker similarity competitively with conventional techniques but also provides an interpretable explanation in terms of voice attributes. We strongly believe that Vo-Ve can enhance evaluation schemes across various speech tasks due to its high-level explainability.