Dynamic Lagging for Simultaneous Translation

• URL: http://arxiv.org/abs/2609.05799v1
• Date: Sat, 05 Sep 2026 01:27:19 GMT
• Title: Dynamic Lagging for Simultaneous Translation
• Abstract summary: In cascaded simultaneous speech translation, the machine translation system cannot control the read--write schedule of the upstream recognizer.<n>We make a sentence-trained, decoder-only LLM prefix-aware by fine-tuning it on stable prefixes.<n>A single training-free threshold on that confidence is the most effective of the three latency controls we compare.
• Score: 2.6371884203154186
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: In cascaded simultaneous speech translation, the machine translation (MT) system cannot control the read--write schedule of the upstream recognizer: it must decide, from a growing source prefix, how much target text to commit. We make a sentence-trained, decoder-only LLM prefix-aware by fine-tuning it on stable prefixes---the longest prefix that any translation up to the current partial source has shared with the model's own full-source output---mixed with full-sentence pairs, and prompt it through a single force-decode turn that carries the committed target forward as more source arrives, making the system flicker-free by construction. We fine-tune Qwen3-8B for EN to DE, JA, ZH, simulating the source stream with reference-transcript prefixes. Prefix finetuning preserves full-sentence quality while improving worst-position chunk quality, and it improves calibration of token-level commit confidence, reducing expected calibration error (ECE) on early source prefixes against a stable-prefix oracle. A single training-free threshold on that confidence is the most effective of the three latency controls we compare: it traces a continuous quality--latency frontier that outperforms the discrete wait-$k$ and target-suffix-deletion quality-latency tradeoff mechanisms. The effect holds well on FLEURS, WMT24++, and CoVoST~2 test sets, under both COMET and MetricX.

Related papers

• PRISM: Prior Rectification and Uncertainty-Aware Structure Modeling for Diffusion-Based Text Image Super-Resolution [51.96078493242164]
  PRISM is a single-step diffusion-based Text-SR framework.<n> PRISM achieves state-of-the-art performance with millisecond-level inference.
  arXiv  Detail & Related papers  (2026-05-13T05:31:06Z)
• Beyond Scattered Acceptance: Fast and Coherent Inference for DLMs via Longest Stable Prefixes [10.877713536966601]
  Longestahead Prefix (LSP) scheduler is a training-free and model-agnostic inference paradigm based on monolithic prefix absorption.<n>LSP evaluates token stability via a single forward pass, dynamically identifies a contiguous left-aligned block of stable predictions.<n>It snaps its boundary to natural linguistic or structural acceptances before an atomic commitment.
  arXiv  Detail & Related papers  (2026-03-05T18:25:26Z)
• EffiReasonTrans: RL-Optimized Reasoning for Code Translation [48.862761776440635]
  EffiReasonTrans is a training framework designed to improve translation accuracy while balancing inference latency.<n>EffiReasonTrans consistently improves translation accuracy (up to +49.2% CA and +27.8% CodeBLEU compared to the base model) while reducing the number of generated tokens.
  arXiv  Detail & Related papers  (2025-10-21T17:55:39Z)
• Language Model is a Branch Predictor for Simultaneous Machine Translation [73.82754138171587]
  We propose incorporating branch prediction techniques in SiMT tasks to reduce translation latency. We utilize a language model as a branch predictor to predict potential branch directions. When the actual source word deviates from the predicted source word, we use the real source word to decode the output again, replacing the predicted output.
  arXiv  Detail & Related papers  (2023-12-22T07:32:47Z)
• CBSiMT: Mitigating Hallucination in Simultaneous Machine Translation with Weighted Prefix-to-Prefix Training [13.462260072313894]
  Simultaneous machine translation (SiMT) is a challenging task that requires starting translation before the full source sentence is available. Prefix-to- framework is often applied to SiMT, which learns to predict target tokens using only a partial source prefix. We propose a Confidence-Based Simultaneous Machine Translation framework, which uses model confidence to perceive hallucination tokens.
  arXiv  Detail & Related papers  (2023-11-07T02:44:45Z)
• Incremental Blockwise Beam Search for Simultaneous Speech Translation with Controllable Quality-Latency Tradeoff [49.75167556773752]
  Blockwise self-attentional encoder models have emerged as one promising end-to-end approach to simultaneous speech translation. We propose a modified incremental blockwise beam search incorporating local agreement or hold-$n$ policies for quality-latency control.
  arXiv  Detail & Related papers  (2023-09-20T14:59:06Z)
• Non-Autoregressive Neural Machine Translation: A Call for Clarity [3.1447111126465]
  We take a step back and revisit several techniques that have been proposed for improving non-autoregressive translation models. We provide novel insights for establishing strong baselines using length prediction or CTC-based architecture variants. We contribute standardized BLEU, chrF++, and TER scores using sacreBLEU on four translation tasks.
  arXiv  Detail & Related papers  (2022-05-21T12:15:22Z)
• Anticipation-free Training for Simultaneous Translation [70.85761141178597]
  Simultaneous translation (SimulMT) speeds up the translation process by starting to translate before the source sentence is completely available. Existing methods increase latency or introduce adaptive read-write policies for SimulMT models to handle local reordering and improve translation quality. We propose a new framework that decomposes the translation process into the monotonic translation step and the reordering step.
  arXiv  Detail & Related papers  (2022-01-30T16:29:37Z)
• Source and Target Bidirectional Knowledge Distillation for End-to-end Speech Translation [88.78138830698173]
  We focus on sequence-level knowledge distillation (SeqKD) from external text-based NMT models. We train a bilingual E2E-ST model to predict paraphrased transcriptions as an auxiliary task with a single decoder.
  arXiv  Detail & Related papers  (2021-04-13T19:00:51Z)
• Composed Fine-Tuning: Freezing Pre-Trained Denoising Autoencoders for Improved Generalization [93.95299500688286]
  We focus on prediction problems with structured outputs subject to output validity constraints. We propose composed fine-tuning, which fine-tunes a predictor composed with the pre-trained denoiser. For two-layer ReLU networks, we prove that composed fine-tuning significantly reduces the complexity of the predictor.
  arXiv  Detail & Related papers  (2020-06-29T17:14:35Z)

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.