SqueezeComposer: Temporal Speed-up is A Simple Trick for Long-form Music Composing

2026-03-22

AI Summary

This work addresses the challenges of long-form music generation, which include modeling long-range dependencies and high computational costs. The authors propose a temporal acceleration–deceleration strategy: first generating temporally compressed music sequences (e.g., at 2×–8× speed) efficiently in an accelerated domain using a diffusion model, then restoring the output to its original tempo followed by refinement. This approach reframes long-sequence generation as a hierarchical task and can be seamlessly integrated into existing models. Experimental results demonstrate significant improvements in efficiency and scalability—particularly for full-song accompaniment generation—while preserving high audio quality.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation
• Computer Vision: Diffusion Models for Vision

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Composing coherent long-form music remains a significant challenge due to the complexity of modeling long-range dependencies and the prohibitive memory and computational requirements associated with lengthy audio representations. In this work, we propose a simple yet powerful trick: we assume that AI models can understand and generate time-accelerated (speeded-up) audio at rates such as 2x, 4x, or even 8x. By first generating a high-speed version of the music, we greatly reduce the temporal length and resource requirements, making it feasible to handle long-form music that would otherwise exceed memory or computational limits. The generated audio is then restored to its original speed, recovering the full temporal structure. This temporal speed-up and slow-down strategy naturally follows the principle of hierarchical generation from abstract to detailed content, and can be conveniently applied to existing music generation models to enable long-form music generation. We instantiate this idea in SqueezeComposer, a framework that employs diffusion models for generation in the accelerated domain and refinement in the restored domain. We validate the effectiveness of this approach on two tasks: long-form music generation, which evaluates temporal-wise control (including continuation, completion, and generation from scratch), and whole-song singing accompaniment generation, which evaluates track-wise control. Experimental results demonstrate that our simple temporal speed-up trick enables efficient, scalable, and high-quality long-form music generation. Audio samples are available at https://SqueezeComposer.github.io/.

Problem

• long-form music generation
• long-range dependencies
• computational efficiency
• temporal modeling
• music coherence

Innovation

• temporal speed-up
• long-form music generation
• diffusion models
• hierarchical generation
• audio acceleration