Video-to-Audio Generation with Hidden Alignment

TL;DR

This paper explores methods for generating synchronized audio from video inputs, focusing on vision encoders, auxiliary embeddings, and data augmentation, achieving state-of-the-art results in video-to-audio alignment.

Contribution

It introduces a foundational model and comprehensive evaluation pipeline, providing new insights into the effects of different encoders, embeddings, and augmentation techniques for improved video-to-audio generation.

Findings

State-of-the-art video-audio synchronization performance

Impact of data augmentation on generation quality

Effectiveness of various vision encoders and embeddings

Abstract

Generating semantically and temporally aligned audio content in accordance with video input has become a focal point for researchers, particularly following the remarkable breakthrough in text-to-video generation. In this work, we aim to offer insights into the video-to-audio generation paradigm, focusing on three crucial aspects: vision encoders, auxiliary embeddings, and data augmentation techniques. Beginning with a foundational model built on a simple yet surprisingly effective intuition, we explore various vision encoders and auxiliary embeddings through ablation studies. Employing a comprehensive evaluation pipeline that emphasizes generation quality and video-audio synchronization alignment, we demonstrate that our model exhibits state-of-the-art video-to-audio generation capabilities. Furthermore, we provide critical insights into the impact of different data augmentation methods on enhancing the generation framework's overall capacity. We showcase possibilities to advance the challenge of generating synchronized audio from semantic and temporal perspectives. We hope these insights will serve as a stepping stone toward developing more realistic and accurate audio-visual generation models.