DiEmo-TTS: Disentangled Emotion Representations via Self-Supervised Distillation for Cross-Speaker Emotion Transfer in Text-to-Speech

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cho, Deok-Hyeon, Oh, Hyung-Seok, Kim, Seung-Bin, Lee, Seong-Whan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917020341108736
author Cho, Deok-Hyeon
Oh, Hyung-Seok
Kim, Seung-Bin
Lee, Seong-Whan
author_facet Cho, Deok-Hyeon
Oh, Hyung-Seok
Kim, Seung-Bin
Lee, Seong-Whan
contents Cross-speaker emotion transfer in speech synthesis relies on extracting speaker-independent emotion embeddings for accurate emotion modeling without retaining speaker traits. However, existing timbre compression methods fail to fully separate speaker and emotion characteristics, causing speaker leakage and degraded synthesis quality. To address this, we propose DiEmo-TTS, a self-supervised distillation method to minimize emotional information loss and preserve speaker identity. We introduce cluster-driven sampling and information perturbation to preserve emotion while removing irrelevant factors. To facilitate this process, we propose an emotion clustering and matching approach using emotional attribute prediction and speaker embeddings, enabling generalization to unlabeled data. Additionally, we designed a dual conditioning transformer to integrate style features better. Experimental results confirm the effectiveness of our method in learning speaker-irrelevant emotion embeddings.
format Preprint
id arxiv_https___arxiv_org_abs_2505_19687
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DiEmo-TTS: Disentangled Emotion Representations via Self-Supervised Distillation for Cross-Speaker Emotion Transfer in Text-to-Speech
Cho, Deok-Hyeon
Oh, Hyung-Seok
Kim, Seung-Bin
Lee, Seong-Whan
Sound
Artificial Intelligence
Audio and Speech Processing
Cross-speaker emotion transfer in speech synthesis relies on extracting speaker-independent emotion embeddings for accurate emotion modeling without retaining speaker traits. However, existing timbre compression methods fail to fully separate speaker and emotion characteristics, causing speaker leakage and degraded synthesis quality. To address this, we propose DiEmo-TTS, a self-supervised distillation method to minimize emotional information loss and preserve speaker identity. We introduce cluster-driven sampling and information perturbation to preserve emotion while removing irrelevant factors. To facilitate this process, we propose an emotion clustering and matching approach using emotional attribute prediction and speaker embeddings, enabling generalization to unlabeled data. Additionally, we designed a dual conditioning transformer to integrate style features better. Experimental results confirm the effectiveness of our method in learning speaker-irrelevant emotion embeddings.
title DiEmo-TTS: Disentangled Emotion Representations via Self-Supervised Distillation for Cross-Speaker Emotion Transfer in Text-to-Speech
topic Sound
Artificial Intelligence
Audio and Speech Processing
url https://arxiv.org/abs/2505.19687