Qwen-Music Technical Report

2026-07-13

AI Summary

This work addresses the challenge of generating high-quality, high-fidelity music with full vocal performances by proposing an end-to-end system capable of text-to-song synthesis and stylized cover singing. The system integrates an audio semantic tokenizer, a large language model, and a generative renderer, featuring a novel melody-token-based Chain-of-Thought (Melody-CoT) mechanism that plans global melodic structure prior to generation. It leverages 25 Hz single-codebook tokenization, autoregressive semantic modeling, and stereo rendering. Through quality-aware pretraining and progressive post-training—comprising supervised initialization, offline DPO, and online GSPO—the method achieves state-of-the-art performance on 13 out of 16 objective metrics across 600 bilingual prompts, surpasses leading commercial systems in expert evaluations, and demonstrates superior fidelity in preserving reference melodies for cover generation.

Abstract

In this report, we introduce Qwen-Music, a powerful music generation model capable of producing highly musical and high-fidelity songs with complete vocal singing. Qwen-Music supports two core tasks: Text to Music Generation, which create entirely new songs from text descriptions, lyrics, and musical attributes, and Cover Song Generation, which reinterprets existing songs with different styles and vocal characteristics. Architecturally, Qwen-Music integrates three core components: Qwen-Music-Tokenizer, Qwen-Music-LLM, and Qwen-Music-Render. Qwen-Music-Tokenizer compresses audio into a 25 Hz single-codebook stream of Music Semantic Tokens that preserve semantic and melodic information for LLM prediction. Based on these tokens, Qwen-Music-LLM performs autoregressive music semantic modeling, with a key novelty being a melody-token-based chain-of-thought (Melody-CoT) mechanism that plans melodies before full-song generation, improving creativity, musicality, structural coherence, and reference-audio-based melody cloning. To overcome the fidelity limitations of discrete semantic tokens, Qwen-Music-Render performs generative stereo rendering, enriching acoustic details and producing high-fidelity stereo waveforms. Finally, we train Qwen-Music-LLM on more than 5 million hours of multilingual music data covering hundreds of languages. We first apply quality-aware pre-training curriculum, then use progressive post-training, comprising supervised initialization, offline DPO, and online GSPO, to further improve musicality and instruction-following ability. Across 600 Chinese and English prompts, Qwen-Music achieves state-of-the-art results in 13 of 16 objective musicality and audio-quality metrics. Professional evaluators also prefer Qwen-Music over leading proprietary systems. For cover song generation, Qwen-Music preserves reference melodies more accurately than leading proprietary systems.

Problem

• music generation
• text-to-music
• cover song generation
• high-fidelity audio
• vocal singing

Innovation

• Melody-CoT
• Music Semantic Tokens
• Generative Stereo Rendering
• Chain-of-Thought for Music
• High-Fidelity Singing Synthesis