# Using joint training speaker encoder with consistency loss to achieve cross-lingual voice conversion and expressive voice conversion

arXiv: https://arxiv.org/abs/2307.00393 · HTML: https://mlsanity.com/paper/2307.00393/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 1 Jul 2023.

## Built on by (most cited first, 5 of 5)

- [MulliVC: Multi-lingual Voice Conversion With Cycle Consistency](https://mlsanity.com/paper/2408.04708/) (2024)
- [Towards Better Disentanglement in Non-Autoregressive Zero-Shot Expressive Voice Conversion](https://mlsanity.com/paper/2506.04013/) (2025)
- [NaturalVoices: A Large-Scale, Spontaneous and Emotional Podcast Dataset for Voice Conversion](https://mlsanity.com/paper/2511.00256/) (2025)
- [As Good as It KAN Get: High-Fidelity Audio Representation](https://mlsanity.com/paper/2503.02585/) (2025)
- [Lightweight and perceptually-guided voice conversion for electro-laryngeal speech](https://mlsanity.com/paper/2601.03892/) (2026)

## Abstract (arXiv)

Voice conversion systems have made significant advancements in terms of naturalness and similarity in common voice conversion tasks. However, their performance in more complex tasks such as cross-lingual voice conversion and expressive voice conversion remains imperfect. In this study, we propose a novel approach that combines a jointly trained speaker encoder and content features extracted from the cross-lingual speech recognition model Whisper to achieve high-quality cross-lingual voice conversion. Additionally, we introduce a speaker consistency loss to the joint encoder, which improves the similarity between the converted speech and the reference speech. To further explore the capabilities of the joint speaker encoder, we use the phonetic posteriorgram as the content feature, which enables the model to effectively reproduce both the speaker characteristics and the emotional aspects of the reference speech.