StressTransfer: Stress-Aware Speech-to-Speech Translation with Emphasis Preservation

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Chen, Xi, Song, Yuchen, Nakamura, Satoshi
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866918160941187072
author Chen, Xi
Song, Yuchen
Nakamura, Satoshi
author_facet Chen, Xi
Song, Yuchen
Nakamura, Satoshi
contents We propose a stress-aware speech-to-speech translation (S2ST) system that preserves word-level emphasis by leveraging LLMs for cross-lingual emphasis conversion. Our method translates source-language stress into target-language tags that guide a controllable TTS model. To overcome data scarcity, we developed a pipeline to automatically generate aligned training data and introduce the "LLM-as-Judge" for evaluation. Experiments show our approach substantially outperforms baselines in preserving emphasis while maintaining comparable translation quality, speaker intent, and naturalness. Our work highlights the importance of prosody in translation and provides an effective, data-efficient solution for preserving paralinguistic cues in S2ST.
format Preprint
id arxiv_https___arxiv_org_abs_2510_13194
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle StressTransfer: Stress-Aware Speech-to-Speech Translation with Emphasis Preservation
Chen, Xi
Song, Yuchen
Nakamura, Satoshi
Computation and Language
Artificial Intelligence
We propose a stress-aware speech-to-speech translation (S2ST) system that preserves word-level emphasis by leveraging LLMs for cross-lingual emphasis conversion. Our method translates source-language stress into target-language tags that guide a controllable TTS model. To overcome data scarcity, we developed a pipeline to automatically generate aligned training data and introduce the "LLM-as-Judge" for evaluation. Experiments show our approach substantially outperforms baselines in preserving emphasis while maintaining comparable translation quality, speaker intent, and naturalness. Our work highlights the importance of prosody in translation and provides an effective, data-efficient solution for preserving paralinguistic cues in S2ST.
title StressTransfer: Stress-Aware Speech-to-Speech Translation with Emphasis Preservation
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2510.13194