English

AVTok: 1D Unified Tokenization for Holistic Audio-Video Generation

Computer Vision and Pattern Recognition 2026-06-29 v1 Multimedia Sound Audio and Speech Processing

Abstract

Audio-video generation has recently gained unprecedented research attention, aiming to synthesize high-quality sounding video content with fine-grained synchronization and semantic alignment between the auditory and visual components. The preceding methods predominantly adopt a dual-branch design with separate tokenization and generation modules per modality, neglecting the representation gap while necessitating intensive computational resources for proper training. Inspired by recent advancements in one-dimensional visual tokenization, we present \textbf{AVTok}, a novel unified tokenizer designated for holistic audio-video generation. AVTok features a dual-stream transformer-based architecture with shared encoder-decoder and modal-specific learnable queries to efficiently and effectively encode an audio-video pair into a compact one-dimensional latent representation with a unified codebook. To cope with the heterogeneous information imbalance that hinders AVTok from exploiting aligned audio-visual information, we devise a hierarchical training strategy to progressively realize reconstruction capabilities for each modality. Extensive experiments demonstrate that AVTok excels both in audio-video reconstruction and when integrated into downstream pipelines for audio-to-video, video-to-audio, and class-conditional joint audio-video generation. AVTok paves the way for the challenge of joint audio-video tokenization and provides a potential direction to build unified large multimodal models for audio-video generation.

Keywords

Cite

@article{arxiv.2606.30811,
  title  = {AVTok: 1D Unified Tokenization for Holistic Audio-Video Generation},
  author = {Kien T. Pham and I Chieh Chen and Qifeng Chen and Long Chen},
  journal= {arXiv preprint arXiv:2606.30811},
  year   = {2026}
}

Comments

ECCV 2026