Hear Your Face: Face-based voice conversion with F0 estimation

TL;DR

This paper introduces a face-based voice conversion method that uses facial images to estimate the target speaker's F0, achieving high-quality speech synthesis and accurate voice-face alignment.

Contribution

It presents a novel framework that leverages facial features to perform voice conversion by estimating F0 solely from facial images, a new approach in the field.

Findings

Superior speech generation quality achieved

Effective alignment of facial features with voice characteristics

Accurate tracking of the target speaker's fundamental frequency

Abstract

This paper delves into the emerging field of face-based voice conversion, leveraging the unique relationship between an individual's facial features and their vocal characteristics. We present a novel face-based voice conversion framework that particularly utilizes the average fundamental frequency of the target speaker, derived solely from their facial images. Through extensive analysis, our framework demonstrates superior speech generation quality and the ability to align facial features with voice characteristics, including tracking of the target speaker's fundamental frequency.