Skip to content
Preprint

Breaking the Curse of Multilinguality in Many-to-Many Speech-to-Text Translation via a Resource-Aware Mixture of Speech Encoders

Aug 2026 · 0 citations · 14 references
Computer Science

TL;DR

Empirical analyses show that MoSE improves high-, medium-, and low-resource languages simultaneously, with the largest gains on low-resource speech, thereby breaking the curse of multilinguality without compromising high-resource performance.

Abstract

Multimodal large language models (MLLMs) have achieved significant success in speech-to-text translation (S2TT). However, when processing multilingual speech inputs, a single speech encoder shared across all languages suffers from the curse of multilinguality: languages at different resource levels compete for limited representation capacity, leading to strong high-resource performance but substantial degradation on low-resource speech. To address this problem and improve multilingual consistency, we propose MSRT, a novel framework built around a resource-aware Mixture of Speech Encoders (MoSE). MoSE uses an explicit language router to assign each utterance to an appropriate expert encoder. A frozen expert preserves high-resource language capabilities, while a trainable expert adapts to and specializes in medium- and low-resource languages. We further introduce a five-stage curriculum learning strategy that substantially reduces data dependence, requiring only 10 hours of paired S2TT data per language for effective alignment. We conduct extensive experiments on 45 languages, systematically evaluating all $45 \times 44$ translation directions. Our 4B-parameter model achieves state-of-the-art performance, outperforming substantially larger baselines. Empirical analyses show that MoSE improves high-, medium-, and low-resource languages simultaneously, with the largest gains on low-resource speech, thereby breaking the curse of multilinguality without compromising high-resource performance. To support future multilingual S2TT research, we release our code and models.

View source

Similar papers

Preprint Jul 2026

Evaluating the Effect of Linguistic Relatedness on Cross-Lingual Transfer in Large Multilingual Automatic Speech Recognition

In every setting, pre-adaptation on related auxiliary languages yields no practically meaningful improvements once as little as one hour of target-language data is available, suggesting that relatedness alone may not reliably predict transfer gains in large multilingual ASR, or constitute an effective strategy for extending such models to low-resource languages.

A. Florian, C. Amol, Hope Kerubo Ombaba et al. · 0 citations
Conference Open access 2026

LCMA-SRT: Language-Conditional Mixture-of-Experts Adapters for Joint Multilingual Speech Recognition and Translation

Neural transducers offer an alignment-free framework for speech-to-text modeling, and hierarchical transducer architectures further improve multilingual joint automatic speech recognition (ASR) and speech translation (ST) by stacking a translation-focused encoder on top of an ASR encoder. However, extending hierarchical transducers to multilingual many-to-many settings remains challenging: fully shared models often suffer from negative transfer and unstable target-language generation, while training separate models for each direction is computationally prohibitive. We propose LCMA-SRT (Language-Conditional Mixture-of-Experts Adapters for Speech Recognition and Translation), which augments a hierarchical transducer with language-conditional Mixture-of-Experts (MoE) adapters. A source-conditioned MoE adapter (SRC-MoE) uses source-language embeddings to reduce cross-language interference and improve multilingual ASR. A target-conditioned MoE adapter (TGT-MoE) uses the desired target language to reduce cross-target interference and stabilize target-language generation in many-to-many ST. Experiments on Europarl-ST (9 languages, 72 directions) show that LCMA-SRT improves both ASR and ST within a single joint model, reducing average WER and improving BLEU and COMET over strong hierarchical transducer baselines. We release our code and models at https:

Nan Li, Xiaoyong Guo, Hao Huang et al. · 0 citations
Conference Jul 2026

A Telugu-Adapted Low-Resource Speech-to-Speech Translation System using IndicWhisper, IndicTrans2, and Language-Aware Processing

Speech-to-Speech Translation (S2ST) is a vital technology for facilitating communication across languages, especially in multilingual and resource-limited environments. However, it is still hard to make reliable S2ST systems for low-resource languages like Telugu because there aren’t many parallel speech resources, the morphology is complex, the speakers are different, and the real world is noisy. This paper introduces a Telugu-adapted cascaded S2ST framework that incorporates IndicWhisper for automatic speech recognition, IndicTrans2 for machine translation, and text-to-speech synthesis for generating spoken output. The primary contribution of this work is the integration of Telugu-aware preprocessing and post-processing techniques, encompassing audio normalization, noise reduction, sentence-level segmentation, Telugu script normalization, numerical normalization, and named-entity preservation. These language-aware parts are meant to make transcription more reliable and translation more accurate when there aren’t many low resources.We test the proposed system with benchmark and custom datasets, such as TeluguST-46, the BhashaAnuvaad subset, and recordings from the real world that are noisy. The full system gets a BLEU score of 0.62, a WER of 0.21, and a chrF score of 0.89, which is better than a plain cascade baseline. Ablation analysis further substantiates that both preprocessing and post-processing significantly influence the final performance. The research illustrates that language-aware adaptation enhances cascaded S2ST systems for Telugu and other low-resource languages.

Navya Guggilam, Vijaya Babu Panthagani, Bollimunta Ankammarao et al. · 0 citations
Preprint Aug 2026

VoxSumm: A Multilingual Corpus of Long-Form Spoken News for Joint Summarization and Translation

This work formalizes joint speech summarization and translation (JSumT), the generation of a succinct, faithful target-language summary directly from a long spoken document in a source language, and establishes a foundation for developing and evaluating multilingual systems capable of jointly interpreting, compressing, and translating long-form speech.

Yejin Jeon, Marie Maltais, Virginia Ceccatelli et al. · 0 citations
Open access Aug 2026

Audio-guided articulatory distillation for multilingual visual speech recognition with large language model decoding

Visual Speech Recognition (VSR) remains challenging in multilingual and low-resource settings due to visual ambiguity, limited annotated data, and weak cross-lingual generalization. This paper introduces an audio-guided distillation framework for multilingual VSR that exploits synchronized audio-visual speech during training while preserving visual-only inference. The proposed architecture consists of an audio-visual teacher that learns articulation-aware representations from aligned acoustic and video streams, and a visual-only student trained to approximate the teacher through representation- and decoder-level distillation. To improve transcription under ambiguous visual evidence, continuous speech representations are projected into the embedding space of a pretrained large language model through a lightweight adaptation module, enabling language-conditioned decoding without full LLM retraining. We further introduce RoVSR-II, an extended Romanian in-the-wild VSR corpus comprising approximately 250 h of audiovisual speech, designed to support evaluation in an underrepresented language. Experiments on mTEDx demonstrate consistent improvements over existing multilingual VSR methods across Latin-script languages in terms of Word Error Rate (WER) and Character Error Rate (CER). Additional evaluation on RoVSR-II shows that the proposed model supports zero-shot transfer to Romanian and substantially reduces both WER and CER after parameter-efficient supervised adaptation. Ablation results further confirm the contribution of LLM decoding, teacher initialization, and audio-guided distillation to visual-only recognition performance. © 2017 Elsevier Inc. All rights reserved.

B. Mocanu, Ruxandra Tapu · 0 citations
Preprint Aug 2026

Confucius4-TTS: Transcript-Free Cross-Lingual Zero-Shot TTS with a Learnable Speaker Encoder

Recent advances in zero-shot text-to-speech (TTS) have substantially improved speech quality and voice cloning fidelity. However, many zero-shot TTS systems still depend on audio prompt transcripts at inference time. This dependency limits cross-lingual voice cloning, since in-the-wild reference audio is often untranscribed. In this technical report, we present Confucius4-TTS, a multilingual zero-shot TTS system that supports 14 languages and performs both intra-lingual and cross-lingual reference cloning without requiring transcripts of audio prompts. Confucius4-TTS follows a two-stage architecture, consisting of text-to-semantic (T2S) and semantic-to-acoustic (S2A) modules. The LLM-based T2S module uses a learnable speaker encoder to extract timbre features from self-supervised speech representations, and the conditional flow-matching S2A module converts the predicted semantic tokens into mel-spectrograms. The same model also supports continuation cloning when a reference transcript is available. Confucius4-TTS is trained on large-scale multilingual speech data. It achieves high intelligibility and speaker similarity on public benchmarks. On the CV3-Eval cross-lingual benchmark, Confucius4-TTS obtains an average WER of 3.73% across six directions. On our internal cross-lingual set, it achieves the best average overall rank in human evaluation among recent open-source and commercial systems. We release code, model checkpoints, and demos at https://github.com/netease-youdao/Confucius4-TTS.

Huaxuan Wang, Huimin Wang, Ruiyu Zhang et al. · 0 citations