SimulU: Training-free Policy for Long-form Simultaneous Speech-to-Speech Translation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Djanibekov, Amirbek, Bentivogli, Luisa, Negri, Matteo, Papi, Sara
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910057231286272
author Djanibekov, Amirbek
Bentivogli, Luisa
Negri, Matteo
Papi, Sara
author_facet Djanibekov, Amirbek
Bentivogli, Luisa
Negri, Matteo
Papi, Sara
contents Simultaneous speech-to-speech translation (SimulS2S) is essential for real-time multilingual communication, with increasing integration into meeting and streaming platforms. Despite this, SimulS2S remains underexplored in research, where current solutions often rely on resource-intensive training procedures and operate on short-form, pre-segmented utterances, failing to generalize to continuous speech. To bridge this gap, we propose SimulU, the first training-free policy for long-form SimulS2S. SimulU adopts history management and speech output selection strategies that exploit cross-attention in pre-trained end-to-end models to regulate both input history and output generation. Evaluations on MuST-C across 8 languages show that SimulU achieves a better or comparable quality-latency trade-off against strong cascaded models. By eliminating the need for ad-hoc training, SimulU offers a promising path to end-to-end SimulS2S in realistic, long-form scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2603_16924
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SimulU: Training-free Policy for Long-form Simultaneous Speech-to-Speech Translation
Djanibekov, Amirbek
Bentivogli, Luisa
Negri, Matteo
Papi, Sara
Audio and Speech Processing
Artificial Intelligence
Computation and Language
Simultaneous speech-to-speech translation (SimulS2S) is essential for real-time multilingual communication, with increasing integration into meeting and streaming platforms. Despite this, SimulS2S remains underexplored in research, where current solutions often rely on resource-intensive training procedures and operate on short-form, pre-segmented utterances, failing to generalize to continuous speech. To bridge this gap, we propose SimulU, the first training-free policy for long-form SimulS2S. SimulU adopts history management and speech output selection strategies that exploit cross-attention in pre-trained end-to-end models to regulate both input history and output generation. Evaluations on MuST-C across 8 languages show that SimulU achieves a better or comparable quality-latency trade-off against strong cascaded models. By eliminating the need for ad-hoc training, SimulU offers a promising path to end-to-end SimulS2S in realistic, long-form scenarios.
title SimulU: Training-free Policy for Long-form Simultaneous Speech-to-Speech Translation
topic Audio and Speech Processing
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2603.16924