English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Video to sound generation aims to generate realistic and natural sound givena video input. However, previous video-to-sound generation methods can onlygenerate a random or average timbre without any controls or specializations ofthe generated sound timbre, leading to the problem that people cannot obtainthe desired timbre under these methods sometimes. In this paper, we pose thetask of generating sound with a specific timbre given a video input and areference audio sample. To solve this task, we disentangle each target soundaudio into three components: temporal information, acoustic information, andbackground information. We first use three encoders to encode these componentsrespectively: 1) a temporal encoder to encode temporal information, which isfed with video frames since the input video shares the same temporalinformation as the original audio; 2) an acoustic encoder to encode timbreinformation, which takes the original audio as input and discards its temporalinformation by a temporal-corrupting operation; and 3) a background encoder toencode the residual or background sound, which uses the background part of theoriginal audio as input. To make the generated result achieve better qualityand temporal alignment, we also adopt a mel discriminator and a temporaldiscriminator for the adversarial training. Our experimental results on the VASdataset demonstrate that our method can generate high-quality audio sampleswith good synchronization with events in video and high timbre similarity withthe reference audio.

VarietySound: Timbre-Controllable Video to Sound Generation via Unsupervised Information Disentanglement