YuE2 Music
Next-Generation Explainable Music Intelligence

Stop Playing Prompt Roulette:
Why Serious Producers Choose White-Box AI

Black-box audio models jump straight from text prompts to raw waveforms. If one bar sounds out of key, you are forced to re-roll the entire track. White-box AI decouples musical composition into editable symbolic notation, giving producers full authorship and precision control.

Deconstructing the YuE2 Two-Stage Pipeline

Mirroring professional studio production: symbolic composition first, acoustic mastering second

STAGE 01Symbolic Planning

Symbolic Score Planning

AI establishes the core musical theory: key signature, tempo grid, melody pitch contours, and harmonic chord changes in clean ABC 2.1 notation.

X: 1
T: Midnight Highway
M: 4/4 | Q: 1/4=124 | K: Dm
|: "Dm" D2 F2 A2 d2 | "Bb" d2 B2 G2 D2 :|
Human-in-the-loop: Edit anytime before rendering
STAGE 02Neural Acoustic Diffusion

48kHz Acoustic Synthesis

YuE2-3B neural diffusion model synthesizes the approved symbolic score into pristine 48kHz stereo master audio with zero out-of-tune hallucinations.

Format: 48,000 Hz Stereo FLAC
Dynamic Range: > 96 dB
Latency: Instant Score Synthesis
Hi-Fi master output with full commercial license

Black-Box Roulette vs White-Box Precision

Edit Melody or Lyrics

黑盒模型 (Suno):必须重投点数重新生成整首歌,原曲好听的段落荡然无存。
白盒模型 (YuE2):直接在乐谱或歌词输入框定位对应行修改,保持其他小节原样。

DAW Studio Integration

黑盒模型 (Suno):只有一个平铺的双声道音频文件,无法替换音色或二次编曲。
白盒模型 (YuE2):一键导出多轨 MIDI、五线谱与无损音频,与 Ableton/FL 无缝衔接。

Cost of Experimentation

黑盒模型 (Suno):一首歌反复尝试数十次,几十美金迅速消耗殆尽。
白盒模型 (YuE2):在乐谱模式零成本实时合成试听,满意后再执行 48kHz 母带渲染。

Take Full Control of Your Music

Experience white-box AI music production in your browser with 30 free welcome credits.