LTX-2: Efficient Joint Audio-Visual Foundation Model

TL;DR

LTX-2 is a novel open-source audiovisual foundation model that generates high-quality, synchronized video and audio content from text prompts, advancing the integration of semantic, emotional, and atmospheric cues in AI-generated media.

Contribution

It introduces a dual-stream transformer architecture with cross-attention and a modality-aware guidance mechanism for efficient, high-quality audiovisual generation from text prompts.

Findings

Achieves state-of-the-art audiovisual quality among open-source models.

Produces coherent audio tracks matching scene context and emotion.

Operates with lower computational cost compared to proprietary systems.

Abstract

Recent text-to-video diffusion models can generate compelling video sequences, yet they remain silent -- missing the semantic, emotional, and atmospheric cues that audio provides. We introduce LTX-2, an open-source foundational model capable of generating high-quality, temporally synchronized audiovisual content in a unified manner. LTX-2 consists of an asymmetric dual-stream transformer with a 14B-parameter video stream and a 5B-parameter audio stream, coupled through bidirectional audio-video cross-attention layers with temporal positional embeddings and cross-modality AdaLN for shared timestep conditioning. This architecture enables efficient training and inference of a unified audiovisual model while allocating more capacity for video generation than audio generation. We employ a multilingual text encoder for broader prompt understanding and introduce a modality-aware classifier-free guidance (modality-CFG) mechanism for improved audiovisual alignment and controllability. Beyond generating speech, LTX-2 produces rich, coherent audio tracks that follow the characters, environment, style, and emotion of each scene -- complete with natural background and foley elements. In our evaluations, the model achieves state-of-the-art audiovisual quality and prompt adherence among open-source systems, while delivering results comparable to proprietary models at a fraction of their computational cost and inference time. All model weights and code are publicly released.