01
It writes the score first, then performs it
Symbolic planning gives the model a musical plan before audio synthesis. In the official workflow, melody and chord symbols can be inspected between those stages.
YuE2: Write a Song, Then Edit the Score It Wrote
YuE2 is an open-weight music model that turns your lyrics and a style prompt into a full song with vocals. Its official workflow also gives you an editable score. Create audio in this browser studio; explore score editing in the model workflow below.
Create a Song With YuE2
Add your lyrics, describe the musical direction, and generate a complete song in the browser. You can also include reference audio to guide a new rendition.
5 free credits to start. Sign in to generate your first song.
What is YuE2?
YuE2 is an open music generation model that turns written lyrics and a style description into a complete song with vocals and accompaniment, by first composing an editable musical score.
YuE2 comes from M·A·P (Multimodal Art Projection). The generation model is called YuE2-3B. You will also see it written as YuE2 AI, which refers to the same music model.
YuE2 adds a step called symbolic planning: it writes melody and chords as a readable score in ABC notation before rendering audio. In the official model workflow, you can inspect that score, change a chord and render another version. That middle step makes the composition part of the workflow, rather than leaving you with audio alone.
This site provides an online studio for lyrics, style prompts and reference audio. The editable-score features described below belong to the official model workflow; this studio does not provide a score editor or score downloads. Source: YuE2 model card.
YuE2 Key Features
The model separates the composition from the performance, so melody and harmony can become explicit inputs in an editing workflow.
02
The score is yours to work on between stages
Change a chord progression or revise a phrase in the ABC score, then generate a new rendition. This is a model-level workflow for people who want to work on the composition directly.
03
Three planning modes
The official API supports melody-and-chord planning, melody-only planning and generation without a symbolic plan. Our browser composer uses the service's generation workflow and does not expose a planning-mode selector.
04
Covers in a different musical setting
The official cover workflow uses a transcribed melody and a new style prompt. In this studio, YuE2 AI Music Cover provides reference-audio generation with lyrics and style.
05
Agent-assisted music editing
An external agent can help revise a score, style and lyrics before the model renders another version. The official editing demonstration follows a song through several musical revisions; an editing agent is not included in this studio.
What Can You Create With YuE2?
Original songs from your words, fresh renditions guided by reference audio, and new songs inspired by your hum.
Original songs from lyrics
You have a verse and a chorus and want to hear them sung. Paste the lyrics, describe an arrangement and listen to the idea as a song.
Covers in a different genre
The official demos include Auld Lang Syne as jazz-funk and Jingle Bells as heavy metal. They illustrate how a familiar tune can be given a different setting. Our online cover page uses a reference recording to guide a new generation.
Turn your hum into a song
Have a melody in mind? Record up to 120 seconds in your browser or upload an audio file. Add lyrics and describe the sound you want, then use your recording to guide a song with vocals and accompaniment.
YuE2 Songs You Can Listen To Right Now
Start with six official YuE2 examples across English, Chinese, Japanese and Spanish, then keep listening through additional covers. Compare the voices, phrasing and arrangements to find a direction for your own song.
Cyber Metal
Y2YuE2
Nu-Disco
Y2YuE2
Jazz-Funk
Y2YuE2
Easy Listening
Y2YuE2
Heartland Rock
Y2YuE2
Flamenco
Y2YuE2
Dancing With My Eyes Closed
Raymond
Porch Light On
kealix
Together in Respect
evalena
A new world
EpicGarage
This Is My World ( V6 Edition )
Zombie Robby
1nes and zer0s
INJURY
Stay or Leave
M.F[T.O] Beats
No More...
Aldo NX
pick [n]one
MC Escher
deadstar
madebyanubis
Who is YuE2 Built For?
YuE2 suits people who care what the song is made of, not only how the first take sounds. Different workflows suit different starting points.
For songwriters who have the words but not the band
Paste the lyrics, describe the arrangement and listen. The online studio gives you a way to explore a vocal demo before planning a recording session. Review the phrasing and rewrite the words or musical direction for another attempt.
For producers who want to keep working
The official model's score workflow lets you inspect melody and harmony, revise the composition and render another version. Use that workflow when the notes themselves matter; use this studio when you want an audio draft from lyrics, style and a reference.
For developers and researchers who need an open model
YuE2 publishes downloadable weights, evaluation material and separate planning and synthesis interfaces. These give you a way to examine the model and build a workflow around its stages. Explore the model review.
YuE2 vs Suno v6: the Published Benchmark
The YuE2 team's September 12, 2026 comparison reports strengths and tradeoffs across 192 WildSongBench prompts. These are model evaluations, not guarantees for this online service.
| Measure | YuE2 | Suno v6 |
|---|---|---|
| Musicality | 5.9075 | 5.6558 |
| SongBench average | 6.7316 | 6.5562 |
| Prompt adherence (Q3O) | 4.6819 | 4.6258 |
| Pronunciation errors (PER, lower is better) | 8.44% | 7.58% |
| Text alignment (AllMusicCaps) | 0.4054 | 0.4305 |
In this reported comparison, YuE2 scores higher on musicality and SongBench average, while Suno v6 has fewer pronunciation errors and higher AllMusicCaps text alignment. The YuE2 row here is the standard result, not the separate best-of-8 result.
We also ran six songs through both platforms from the same lyrics, and each pair states how closely the two briefs were matched before we say anything about the result. Hear YuE2 and Suno V6 side by side.
Why Choose YuE2 Over Another AI Music Generator
The useful distinction is whether you want a quick audio draft or a workflow that lets you work on the composition itself.
Choose YuE2 when
- You want to hear your writing. The browser studio turns lyrics and a style brief into a song you can assess.
- You want to inspect a composition. The official model workflow exposes an editable score between planning and synthesis.
- You are evaluating open music models. Downloadable weights and published evaluation material let you investigate the model directly.
Consider the limits before you start
- Pronunciation needs review. The published comparison shows a higher phoneme error rate than Suno v6. Listen closely to the words in your own output.
- Score editing needs a separate workflow. This studio does not expose note-level editing or the model's planning controls.
- Every rendition can vary. A prompt describes a direction; it is not a promise of an exact arrangement or an unchanged performance.
How To Use YuE2 in Four Steps
Bring your lyrics, set the style, choose an audio format and follow the generation. These four steps show the browser studio workflow.
1. Add your lyrics
Select Text to Music and paste the words you want performed. Label sections such as Verse, Chorus and Bridge, and leave space between them. If a recording is part of your brief, choose Music to Music and upload a reference as well.
2. Set the musical style
Describe the genre, vocal character, main instruments and mood. Start with a preset or write a focused brief, such as warm indie pop with acoustic guitar, piano and gentle drums. Keep arrangement instructions in Style, separate from the lyrics.
3. Choose a format and generate
Choose MP3, WAV or FLAC, sign in and check your credits before selecting Generate. Text to Music uses 3 credits; Music to Music uses 4. Each new generation uses credits.
4. Follow the progress and listen
The estimated progress bar moves to 95% over about 60 seconds, then waits for the actual result. It is a visual estimate, not a measurement of model completion. When your song is ready, play it on the page and decide what to try next. Downloads are available with paid plans.
For melody and chord editing outside this studio, see the score-editing experiments in our YuE2 review.
YuE2 Pricing
Start with 5 free credits. Text to Music uses 3 credits per generation; Music to Music uses 4. Paid packs are one-time purchases, with no monthly subscription. The Free plan includes generation and listening; paid plans include downloads.
Starter
$9.9
One-time · 99 credits
$0.10 / credit
- 33 songs
- 33 music generations
- 99 lyrics generations
- 2 tasks at a time
- Commercial License Included
- AI Music Generator
- AI Lyrics Generator
- Remix & Edit Song
- Advanced Music Analyzer
- AI Singing Voice Model Training
- Music Download
- Email Support
Basic
$29.9
One-time · 370 credits
$0.08 / credit (approx.)
- 123 songs
- 123 music generations
- 5 tasks at a time
- Commercial License Included
- AI Music Generator
- AI Lyrics Generator
- Remix & Edit Song
- Advanced Music Analyzer
- AI Singing Voice Model Training
- Unlimited Download
- 24/7 email support
Plus
Most popular$49.9
One-time · 700 credits
$0.07 / credit (approx.)
- 233 songs
- 233 music generations
- 10 tasks at a time
- Commercial License Included
- AI Music Generator
- AI Lyrics Generator
- Remix & Edit Song
- Advanced Music Analyzer
- AI Singing Voice Model Training
- Unlimited Download
- 24/7 email support
Frequently Asked Questions About YuE2
What is YuE2?
YuE2 is an open music generation model from M-A-P. It turns lyrics and a style prompt into songs with vocals and accompaniment, with a symbolic planning workflow that can make melody and chords editable before audio rendering.
How much does YuE2 cost here?
The Free plan includes 5 credits. One-time packs start at $9.90 for 99 credits; Basic is $29.90 for 370 and Plus is $49.90 for 700. Text to Music uses 3 credits and Music to Music uses 4. Paid plans include downloads.
How does YuE2 compare to Suno v6?
In the 12 September 2026 comparison, standard YuE2 scores 5.9075 on musicality against 5.6558 for Suno V6. Suno V6 has fewer pronunciation errors: 7.58% against 8.44%. These reported results do not guarantee individual studio outputs. The YuE2 vs Suno V6 page walks through scores, editing, credits and downloads.
Can I run YuE2 locally or in ComfyUI?
The official local quick start specifies Linux, Python 3.10+ and a 24 GB NVIDIA GPU with BF16 support. Community ComfyUI workflows are separate integrations. This online studio does not require either installation.
What languages can I hear in the YuE2 demos?
The six official examples on this homepage include English, Chinese, Japanese and Spanish. They demonstrate the model rather than customer work from this studio. Pronunciation and phrasing should be reviewed for each generated song.
Can YuE2 cover an existing song?
The official model supports covers guided by a transcribed score. Here, Music to Music accepts one reference recording, lyrics and a style prompt to guide another rendition. The studio does not expose a score editor or guarantee an unchanged melody.
What is symbolic planning?
Symbolic planning is the stage where YuE2 writes a melody and chords before rendering audio. The official workflow supports full planning, melody-only planning and direct generation. This browser composer does not expose these planning settings or editable score exports.
What do I need to write a song with YuE2?
Start with lyrics and a description of the sound. Organize the words into sections and put musical directions in Style. Sign in to generate online. Choose MP3, WAV or FLAC and listen on the page; downloads require a paid plan.
Start a Song with YuE2
Bring the words you want to hear. Choose a musical direction and turn the idea into a song. Start with 5 free credits.