VITA-1.5: Towards GPT-4o Level Real-Time Vision and Speech Interaction

TL;DR

VITA-1.5 introduces a multi-stage training approach for multimodal large language models, enabling near real-time vision and speech interaction without separate speech modules, advancing multimodal dialogue systems.

Contribution

It presents a novel multi-stage training methodology that integrates visual and speech understanding in LLMs, achieving efficient real-time multimodal interaction.

Findings

Outperforms state-of-the-art on image, video, and speech benchmarks.

Enables speech-to-speech dialogue without separate ASR and TTS modules.

Achieves near real-time vision and speech interaction.

Abstract

Recent Multimodal Large Language Models (MLLMs) have typically focused on integrating visual and textual modalities, with less emphasis placed on the role of speech in enhancing interaction. However, speech plays a crucial role in multimodal dialogue systems, and implementing high-performance in both vision and speech tasks remains a significant challenge due to the fundamental modality differences. In this paper, we propose a carefully designed multi-stage training methodology that progressively trains LLM to understand both visual and speech information, ultimately enabling fluent vision and speech interaction. Our approach not only preserves strong vision-language capacity, but also enables efficient speech-to-speech dialogue capabilities without separate ASR and TTS modules, significantly accelerating multimodal end-to-end response speed. By comparing our method against state-of-the-art counterparts across benchmarks for image, video, and speech tasks, we demonstrate that our model is equipped with both strong visual and speech capabilities, making near real-time vision and speech interaction. Code has been released at https://github.com/VITA-MLLM/VITA.