The goal of this work is to enhance balanced multimodal understanding in audio-visual large language models (AV-LLMs) by addressing modality bias without additional training. In current AV-LLMs, audio and video features are typically processed jointly in the decoder. While this strategy facilitates unified multimodal understanding, it may introduce modality bias, where the model disproportionately relies on a single dominant modality. In this work, we provide an in-depth analysis of this phenomenon and propose Fork-Merge Decoding (FMD) as a targeted solution. FMD is a simple, training-free inference strategy that rectifies this imbalance without architectural modifications. Specifically, FMD isolates audio and visual inputs in the early decoding layers (fork) to enforce modality-specific reasoning, before integrating them for joint interpretation (merge). This two-stage mechanism effectively mitigates the dominance by ensuring balanced contributions from both modalities. Experiments on VideoLLaMA2, video-SALMONN, and Qwen2.5-Omni demonstrate that FMD consistently enhances multimodal understanding across diverse benchmarks.

The goal of this work is to develop a universal approach for aligning subtitles (i.e., spoken language text with corresponding timestamps) to continuous sign language videos. Prior approaches typically rely on end-to-end training tied to a specific language or dataset, which limits their generality. In contrast, our method Segment, Embed, and Align (SEA) provides a single framework that works across multiple languages and domains. SEA leverages two pretrained models: the first to segment a video sequence into individual signs and the second to embed each sign video clip into a shared latent space with text. Alignment is subsequently performed with a lightweight dynamic programming procedure that runs efficiently on CPU within a minute, even for hour-long episodes. SEA is flexible and can adapt to a wide range of scenarios, utilizing resources from small lexicons to large continuous corpora. Experiments on four sign language datasets demonstrate state-of-the-art alignment performance, highlighting the potential of SEA to generate high-quality parallel data for advancing sign language processing.