Sign language translation (SLT) converts sign language video into spoken language text and holds significant promise for improving accessibility and enabling communication between signing and non-signing communities. While large weakly-aligned datasets have enabled pre-training at scale and gloss-free methods have reduced reliance on expert annotation, high-quality parallel sign video-text pairs for fine-tuning remain scarce, limiting generalisation on long-tail vocabulary and unseen constructions. We propose a corpus augmentation approach that requires no additional human annotation, external sign-language video corpora, or generative video models, relying only on the existing gloss-annotated training corpus and an LLM for sentence generation: per-gloss clips are extracted from training videos via CTC forced-alignment, novel gloss-sentence pairs are generated by a corpus-anchored LLM, and synthetic sequences are assembled through random sentence sampling and clip assignment. The resulting synthetic RGB video-text pairs are architecture-agnostic at the downstream training stage and can be consumed directly by RGB-based SLT models, or converted into pose or feature representations by pipelines that derive such inputs from video. Sincan et al. re-evaluated five recent gloss-free methods under strictly identical conditions; the largest verified gain over the GFSLT-VLP baseline was only 0.98 BLEU-4. Our augmentation, applied within the same framework, achieves +2.92 BLEU-4 without any change to architecture or training protocol. We further identify that synthetic data harms vision-language pretraining despite improving its objectives, and that optimising clip transitions for visual smoothness is counter-productive under L2-based criteria; we propose that abrupt boundaries may act as a form of implicit regularisation. Code is available at https://github.com/robizso/slt-datagen.

翻译：手语翻译(SLT)旨在将手语视频转换为口语文本，在提升无障碍交流、促进健听与非健听群体沟通方面具有重要前景。尽管大规模弱对齐数据集推动了预训练发展，且无词汇标注方法减少了对专家标注的依赖，但用于微调的高质量平行手语视频-文本对仍然稀缺，限制了模型对长尾词汇和未见构式的泛化能力。本文提出一种语料库增强方法，无需额外人工标注、外部手语视频语料库或生成式视频模型，仅依赖现有词汇标注训练语料和用于句子生成的大语言模型：通过CTC强制对齐从训练视频中提取逐词汇片段，利用语料库锚定的大语言模型生成新颖的词汇-句子对，并通过随机句子采样与片段分配组装合成序列。生成的合成RGB视频-文本对在后续训练阶段具有架构无关性，可直接被基于RGB的手语翻译模型使用，或通过从视频提取此类输入的流水线转换为姿态或特征表示。Sincan等人在严格相同条件下重新评估了五种近期无词汇标注方法；相较于GFSLT-VLP基线，最大确认增益仅为0.98 BLEU-4。我们在相同框架内应用所提增强方法，在不改变架构或训练协议的情况下实现了+2.92 BLEU-4的提升。我们进一步发现，合成数据虽能改善视觉-语言预训练目标，但会损害其效果；同时，在基于L2的准则下优化片段间的视觉平滑度会产生反效果——我们提出，突变边界可能起到隐式正则化的作用。代码开源于https://github.com/robizso/slt-datagen。