Mutual Forcing: Dual-Mode Self-Evolution for Fast Autoregressive Audio-Video Character Generation

TL;DR

This paper introduces Mutual Forcing, a novel framework for fast autoregressive audio-video generation that enables efficient, high-quality, long-horizon synchronization without relying on bidirectional teachers.

Contribution

It presents a native causal model with integrated multi-step and few-step generation, eliminating the need for complex distillation pipelines and improving training-inference consistency.

Findings

Achieves comparable or better quality with only 4-8 sampling steps compared to 50 steps in baselines.

Supports flexible sequence lengths and reduces training overhead.

Outperforms prior approaches like Self-Forcing in efficiency and quality.

Abstract

In this work, we propose Mutual Forcing, a framework for fast autoregressive audio-video generation with long-horizon audio-video synchronization. Our approach addresses two key challenges: joint audio-video modeling and fast autoregressive generation. To ease joint audio-video optimization, we adopt a two-stage training strategy: we first train uni-modal generators and then couple them into a unified audio-video model for joint training on paired data. For streaming generation, we ask whether a native fast causal audio-video model can be trained directly, instead of following existing streaming distillation pipelines that typically train a bidirectional model first and then convert it into a causal generator through multiple distillation stages. Our answer is Mutual Forcing, which builds directly on native autoregressive model and integrates few-step and multi-step generation within a single weight-shared model, enabling self-distillation and improved training-inference consistency. The multi-step mode improves the few-step mode via self-distillation, while the few-step mode generates historical context during training to improve training-inference consistency; because the two modes share parameters, these two effects reinforce each other within a single model. Compared with prior approaches such as Self-Forcing, Mutual Forcing removes the need for an additional bidirectional teacher model, supports more flexible training sequence lengths, reduces training overhead, and allows the model to improve directly from real paired data rather than a fixed teacher. Experiments show that Mutual Forcing matches or surpasses strong baselines that require around 50 sampling steps while using only 4 to 8 steps, demonstrating substantial advantages in both efficiency and quality. The project page is available at https://mutualforcing.github.io.