Semantically consistent Video-to-Audio Generation using Multimodal Language Large Model

TL;DR

This paper presents SVA, a novel framework that uses multimodal large language models to generate semantically consistent audio, including sound effects and background music, for videos, enhancing immersive viewer experience.

Contribution

The paper introduces a new video-to-audio generation method leveraging multimodal large language models to produce semantically aligned audio from video content.

Findings

Effective generation of semantically consistent audio for videos.

Utilizes multimodal large language models for understanding video semantics.

Demonstrates satisfactory performance through case studies.

Abstract

Existing works have made strides in video generation, but the lack of sound effects (SFX) and background music (BGM) hinders a complete and immersive viewer experience. We introduce a novel semantically consistent v ideo-to-audio generation framework, namely SVA, which automatically generates audio semantically consistent with the given video content. The framework harnesses the power of multimodal large language model (MLLM) to understand video semantics from a key frame and generate creative audio schemes, which are then utilized as prompts for text-to-audio models, resulting in video-to-audio generation with natural language as an interface. We show the satisfactory performance of SVA through case study and discuss the limitations along with the future research direction. The project page is available at https://huiz-a.github.io/audio4video.github.io/.