Video-to-audio synthesis methods are hindered by limitations in training data and textual descriptions, leading to difficulties in capturing fine-grained acoustic details. A new approach, AC-Foley, addresses these challenges by utilizing reference audio to guide the synthesis process, allowing for more accurate and nuanced audio generation. This method leverages acoustic transfer to bridge the gap between visual and auditory information, enabling the creation of more realistic and detailed audio outputs. By bypassing the need for text prompts, AC-Foley mitigates the issue of semantic granularity gaps and textual ambiguity, resulting in improved audio synthesis capabilities1. This development has significant implications for applications such as sound design and audio post-production, where high-quality audio is crucial. The ability to generate accurate and detailed audio from visual information can greatly enhance the overall quality of multimedia productions, so practitioners in these fields should take note of this innovative approach.