EPISODE · Jun 2, 2026 · 26 MIN
Towards Streaming Synchronized Spatial Audio Generation via Autoregressive Diffusion Transformer
from Daily Paper Cast · host Jingwen Liang, Gengyu Wang
🤗 Upvotes: 28 | eess.AS, cs.MM, cs.SD Authors: Ke Lei, Yu Zhang, Changhao Pan, Xueyi Pu, Wenxiang Guo, Ruiqi Li, Zhou Zhao Title: Towards Streaming Synchronized Spatial Audio Generation via Autoregressive Diffusion Transformer Arxiv: http://arxiv.org/abs/2605.30940v1 Abstract: Real-time and accurate spatial audio generation is pivotal for delivering an immersive experience. However, existing spatial audio synthesis technologies are often encumbered by a tradeoff between generation quality and high inference latency, as well as difficulty in capturing precise spatial information from multimodal inputs. To address these challenges, we propose SwanSphere, a unified streaming framework for high-fidelity spatial audio generation from panoramic videos and text prompts. SwanSphere mainly makes the following contributions: 1) We introduce a causal autoregressive diffusion transformer architecture that enables streaming high-quality spatial audio generation. 2) We design a Spatial Video-Audio Contrastive (SVAC) learning strategy to align the video encoder with the acoustic domain, and further employ a multi-objective online direct preference optimization (ODPO) scheme, resulting in strong spatial perception and robust multimodal spatial audio synthesis. 3) To alleviate the current scarcity of spatial audio datasets, we also develop an automated annotation pipeline for generating detailed spatial captions. Experimental results demonstrate that SwanSphere achieves superior performance in both video-to-spatial and text-to-spatial audio generation tasks. Demos can be found at: https://swanaigc.github.io.
What this episode covers
🤗 Upvotes: 28 | eess.AS, cs.MM, cs.SD Authors: Ke Lei, Yu Zhang, Changhao Pan, Xueyi Pu, Wenxiang Guo, Ruiqi Li, Zhou Zhao Title: Towards Streaming Synchronized Spatial Audio Generation via Autoregressive Diffusion Transformer Arxiv: http://arxiv.org/abs/2605.30940v1 Abstract: Real-time and accurate spatial audio generation is pivotal for delivering an immersive experience. However, existing spatial audio synthesis technologies are often encumbered by a tradeoff between generation quality and high inference latency, as well as difficulty in capturing precise spatial information from multimodal inputs. To address these challenges, we propose SwanSphere, a unified streaming framework for high-fidelity spatial audio generation from panoramic videos and text prompts. SwanSphere mainly makes the following contributions: 1) We introduce a causal autoregressive diffusion transformer architecture that enables streaming high-quality spatial audio generation. 2) We design a Spatial Video-Audio Contrastive (SVAC) learning strategy to align the video encoder with the acoustic domain, and further employ a multi-objective online direct preference optimization (ODPO) scheme, resulting in strong spatial perception and robust multimodal spatial audio synthesis. 3) To alleviate the current scarcity of spatial audio datasets, we also develop an automated annotation pipeline for generating detailed spatial captions. Experimental results demonstrate that SwanSphere achieves superior performance in both video-to-spatial and text-to-spatial audio generation tasks. Demos can be found at: https://swanaigc.github.io.
NOW PLAYING
Towards Streaming Synchronized Spatial Audio Generation via Autoregressive Diffusion Transformer
No transcript for this episode yet
Similar Episodes
No similar episodes found.