Efficient and Adaptive Simultaneous Speech Translation with Fully Unidirectional Architecture

• URL: http://arxiv.org/abs/2504.11809v1
• Date: Wed, 16 Apr 2025 06:46:15 GMT
• Title: Efficient and Adaptive Simultaneous Speech Translation with Fully Unidirectional Architecture
• Authors: Biao Fu, Donglei Yu, Minpeng Liao, Chengxi Li, Yidong Chen, Kai Fan, Xiaodong Shi,
• Abstract summary: Simultaneous speech translation (SimulST) produces translations incrementally while processing partial speech input.<n>Existing LLM-based SimulST approaches incur significant computational overhead due to repeated encoding of bidirectional speech encoder.<n>We introduce Efficient and Adaptive Simultaneous Speech Translation (EASiST) with fully unidirectional architecture.
• Score: 14.056534007451763
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Simultaneous speech translation (SimulST) produces translations incrementally while processing partial speech input. Although large language models (LLMs) have showcased strong capabilities in offline translation tasks, applying them to SimulST poses notable challenges. Existing LLM-based SimulST approaches either incur significant computational overhead due to repeated encoding of bidirectional speech encoder, or they depend on a fixed read/write policy, limiting the efficiency and performance. In this work, we introduce Efficient and Adaptive Simultaneous Speech Translation (EASiST) with fully unidirectional architecture, including both speech encoder and LLM. EASiST includes a multi-latency data curation strategy to generate semantically aligned SimulST training samples and redefines SimulST as an interleaved generation task with explicit read/write tokens. To facilitate adaptive inference, we incorporate a lightweight policy head that dynamically predicts read/write actions. Additionally, we employ a multi-stage training strategy to align speech-text modalities and optimize both translation and policy behavior. Experiments on the MuST-C En$\rightarrow$De and En$\rightarrow$Es datasets demonstrate that EASiST offers superior latency-quality trade-offs compared to several strong baselines.

Related papers

• SimulS2S-LLM: Unlocking Simultaneous Inference of Speech LLMs for Speech-to-Speech Translation [14.57248739077317]
  This paper proposes SimulS2S-LLM, which trains speech LLMs offline and employs a test-time policy to guide simultaneous inference. SimulS2S-LLM achieves simultaneous speech-to-speech translation (Simul-S2ST) by predicting discrete output speech tokens and then synthesising output speech using a pre-trained vocoder.
  arXiv  Detail & Related papers  (2025-04-22T01:05:32Z)
• LLMs Can Achieve High-quality Simultaneous Machine Translation as Efficiently as Offline [16.124385656402744]
  Large Language Models (LLMs) perform excellently in offline machine translation even with a simple prompt "Translate the following sentence from [src lang] into [tgt lang]:"<n>We propose a novel paradigm that includes constructing supervised fine-tuning data for simultaneous machine translation (SiMT)<n>Our approach achieves state-of-the-art performance across various SiMT benchmarks, and preserves the original abilities of offline translation.
  arXiv  Detail & Related papers  (2025-04-13T13:45:53Z)
• Speech Translation Refinement using Large Language Models [8.602429274223693]
  This paper investigates how large language models (LLMs) can improve the performance of speech translation by introducing a joint refinement process.<n>Through the joint refinement of speech translation (ST) and automatic speech recognition (ASR) transcription via LLMs, the performance of the ST model is significantly improved.<n> Experimental results on the MuST-C and CoVoST 2 datasets, which include seven translation tasks, demonstrate the effectiveness of the proposed approach.
  arXiv  Detail & Related papers  (2025-01-25T05:32:42Z)
• DeSTA2: Developing Instruction-Following Speech Language Model Without Speech Instruction-Tuning Data [84.01401439030265]
  Recent end-to-end speech language models (SLMs) have expanded upon the capabilities of large language models (LLMs)<n>We present a simple yet effective automatic process for creating speech-text pair data.<n>Our model demonstrates general capabilities for speech-related tasks without the need for speech instruction-tuning data.
  arXiv  Detail & Related papers  (2024-09-30T07:01:21Z)
• Simultaneous Masking, Not Prompting Optimization: A Paradigm Shift in Fine-tuning LLMs for Simultaneous Translation [5.712277386555735]
  Large language models (LLMs) have achieved state-of-the-art performance in various language processing tasks. We propose SimulMask, a new paradigm for fine-tuning LLMs for simultaneous translation. We have observed a significant translation quality improvement compared to state-of-the-art prompting optimization strategies on five language pairs.
  arXiv  Detail & Related papers  (2024-05-16T21:07:42Z)
• Speech Translation with Large Language Models: An Industrial Practice [64.5419534101104]
  We introduce LLM-ST, a novel and effective speech translation model constructed upon a pre-trained large language model (LLM) By integrating the large language model (LLM) with a speech encoder and employing multi-task instruction tuning, LLM-ST can produce accurate timestamped transcriptions and translations. Through rigorous experimentation on English and Chinese datasets, we showcase the exceptional performance of LLM-ST.
  arXiv  Detail & Related papers  (2023-12-21T05:32:49Z)
• Soft Alignment of Modality Space for End-to-end Speech Translation [49.29045524083467]
  End-to-end Speech Translation aims to convert speech into target text within a unified model. The inherent differences between speech and text modalities often impede effective cross-modal and cross-lingual transfer. We introduce Soft Alignment (S-Align), using adversarial training to align the representation spaces of both modalities.
  arXiv  Detail & Related papers  (2023-12-18T06:08:51Z)
• Simultaneous Machine Translation with Large Language Models [51.470478122113356]
  We investigate the possibility of applying Large Language Models to SimulMT tasks. We conducted experiments using the textttLlama2-7b-chat model on nine different languages from the MUST-C dataset. The results show that LLM outperforms dedicated MT models in terms of BLEU and LAAL metrics.
  arXiv  Detail & Related papers  (2023-09-13T04:06:47Z)
• Data-Driven Adaptive Simultaneous Machine Translation [51.01779863078624]
  We propose a novel and efficient training scheme for adaptive SimulMT. Our method outperforms all strong baselines in terms of translation quality and latency.
  arXiv  Detail & Related papers  (2022-04-27T02:40:21Z)
• Exploring Continuous Integrate-and-Fire for Adaptive Simultaneous Speech Translation [75.86581380817464]
  A SimulST system generally includes two components: the pre-decision that aggregates the speech information and the policy that decides to read or write. This paper proposes to model the adaptive policy by adapting the Continuous Integrate-and-Fire (CIF) Compared with monotonic multihead attention (MMA), our method has the advantage of simpler computation, superior quality at low latency, and better generalization to long utterances.
  arXiv  Detail & Related papers  (2022-03-22T23:33:18Z)
• RealTranS: End-to-End Simultaneous Speech Translation with Convolutional Weighted-Shrinking Transformer [33.876412404781846]
  RealTranS is an end-to-end model for simultaneous speech translation. It maps speech features into text space with a weighted-shrinking operation and a semantic encoder. Experiments show that RealTranS with the Wait-K-Stride-N strategy outperforms prior end-to-end models.
  arXiv  Detail & Related papers  (2021-06-09T06:35:46Z)
• SML: a new Semantic Embedding Alignment Transformer for efficient cross-lingual Natural Language Inference [71.57324258813674]
  The ability of Transformers to perform with precision a variety of tasks such as question answering, Natural Language Inference (NLI) or summarising, have enable them to be ranked as one of the best paradigms to address this kind of tasks at present. NLI is one of the best scenarios to test these architectures, due to the knowledge required to understand complex sentences and established a relation between a hypothesis and a premise. In this paper, we propose a new architecture, siamese multilingual transformer, to efficiently align multilingual embeddings for Natural Language Inference.
  arXiv  Detail & Related papers  (2021-03-17T13:23:53Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.