# Temporally-Grounded Language Generation: Towards Real-Time Vision-Language Models

arXiv: https://arxiv.org/abs/2505.11326 · HTML: https://mlsanity.com/paper/2505.11326/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Video understanding with MLLMs, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 17 of its 32 references that are in the map.
- First posted on arXiv 16 May 2025.

## Placed in

- Multimodal LLMs / vision-language models / [Video understanding with MLLMs](https://mlsanity.com/mllm/video-understanding-with-mllms/)

## Builds on (most cited first, 17 of 17)

- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis](https://mlsanity.com/paper/2405.21075/) (2024)
- [EgoSchema: A Diagnostic Benchmark for Very Long-form Video Language Understanding](https://mlsanity.com/paper/2308.09126/) (2023)
- [ActivityNet-QA: A Dataset for Understanding Complex Web Videos via Question Answering](https://mlsanity.com/paper/1906.02467/) (2019)
- [TGIF-QA: Toward Spatio-Temporal Reasoning in Visual Question Answering](https://mlsanity.com/paper/1704.04497/) (2017)
- [WhisperX: Time-Accurate Speech Transcription of Long-Form Audio](https://mlsanity.com/paper/2303.00747/) (2023)
- [VideoLLM-online: Online Video Large Language Model for Streaming Video](https://mlsanity.com/paper/2406.11816/) (2024)
- [Flash-VStream: Memory-Based Real-Time Understanding for Long Video Streams](https://mlsanity.com/paper/2406.08085/) (2024)
- [HoloAssist: an Egocentric Human Interaction Dataset for Interactive AI Assistants in the Real World](https://mlsanity.com/paper/2309.17024/) (2023)
- [SoccerNet-v2: A Dataset and Benchmarks for Holistic Understanding of Broadcast Soccer Videos](https://mlsanity.com/paper/2011.13367/) (2020)
- [Dispider: Enabling Video LLMs with Active Real-Time Interaction via Disentangled Perception, Decision, and Reaction](https://mlsanity.com/paper/2501.03218/) (2025)
- [Streaming Video Understanding and Multi-round Interaction with Memory-enhanced Knowledge](https://mlsanity.com/paper/2501.13468/) (2025)
- [SVBench: A Benchmark with Temporal Multi-Turn Dialogues for Streaming Video Understanding](https://mlsanity.com/paper/2502.10810/) (2025)
- [OmniMMI: A Comprehensive Multi-modal Interaction Benchmark in Streaming Video Contexts](https://mlsanity.com/paper/2503.22952/) (2025)
- [Eliciting In-Context Learning in Vision-Language Models for Videos Through Curated Data Distributional Properties](https://mlsanity.com/paper/2311.17041/) (2023)
- [Can Foundation Models Watch, Talk and Guide You Step by Step to Make a Cake?](https://mlsanity.com/paper/2311.00738/) (2023)
- [What to Say and When to Say it: Live Fitness Coaching as a Testbed for Situated Interaction](https://mlsanity.com/paper/2407.08101/) (2024)

## Abstract (arXiv)

Vision-language models (VLMs) have shown remarkable progress in offline tasks such as image captioning and video question answering. However, real-time interactive environments impose new demands on VLMs, requiring them to generate utterances that are not only semantically accurate but also temporally precise. We identify two core capabilities necessary for such settings---textitperceptual updating and textitcontingency awareness---and propose a new benchmark task, Temporally-Grounded Language Generation (TGLG), to evaluate them. TGLG requires models to generate utterances in response to streaming video such that both content and timing align with dynamic visual input. To support this benchmark, we curate evaluation datasets from sports broadcasting and egocentric human interaction domains, and introduce a new metric, TRACE, to evaluate TGLG by jointly measuring semantic similarity and temporal alignment. Finally, we present Vision-Language Model with Time-Synchronized Interleaving (VLM-TSI), a model that interleaves visual and linguistic tokens in a time-synchronized manner, enabling real-time language generation without relying on turn-based assumptions. Experimental results show that VLM-TSI significantly outperforms a strong baseline, yet overall performance remains modest---highlighting the difficulty of TGLG and motivating further research in real-time VLMs. Our code is available at https://github.com/yukw777/tglg.