YuE2: Frontier AI Music Generation With Editable Scores
YuE2 generates songs with vocals from lyrics, style prompts, and optional reference audio
About YuE2: Frontier AI Music Generation With Editable Scores
YuE2 is an online AI music generator for turning lyrics and musical direction into songs in a browser. The homepage describes it as a studio for creating full songs with vocals and accompaniment from written lyrics, style prompts, and optional reference audio.
The model workflow is built around symbolic planning. YuE2 first writes melody and chords as an editable ABC notation score before rendering audio, while the browser studio focuses on generating audio drafts rather than exposing full note-level score editing.
The site supports Text to Music and Music to Music generation. Text to Music uses 3 credits per generation, Music to Music uses 4 credits, and users can choose MP3, WAV, or FLAC before generating.
Pricing uses one-time credit packs instead of monthly subscriptions. The Free plan includes 5 credits, while paid packs start at $9.90 for 99 credits and include downloads plus a commercial license.
Features
- Text to Music turns lyrics and a style prompt into vocals and accompaniment
- Music to Music accepts reference audio to guide a new generation
- Text to Music costs 3 credits per generation; Music to Music costs 4 credits
- Output format choices listed on the site are MP3, WAV, and FLAC
Use cases
- Create a full song by entering lyrics and defining musical style in the browser
- Modify chord progressions and melody in the score to produce new song versions
- Generate covers by applying reference melodies to new styles and arrangements
- Compare different style prompts by reusing or changing generation settings
Screenshots
Frequently asked questions
Can I edit the music composition itself with YuE2?
Yes, YuE2 produces an editable score in ABC notation between lyric input and audio generation, allowing users to change chords and melody before creating a new rendition.
How do I use my own melody or reference audio in song creation?
You can upload or record an audio file up to 120 seconds as a guide, then add lyrics and style prompts; YuE2 uses this reference to generate a new song version.
Are there any limits on pronunciation accuracy in the generated songs?
The model has a higher phoneme error rate compared to some alternatives, so users should listen carefully to how lyrics are pronounced in the output.
Does the browser studio expose full score editing?
No. The homepage says score editing needs a separate workflow and the studio does not expose note-level editing
0 comments
Sign in to leave a comment.
No comments yet. Say something useful and be the first.