Към всички skills

yue2-music

multimodal-art-projection · Данни и AI

Generate, cover, transcribe, and edit songs with YuE2 and SheetSage2/MERT2. Use for YuE2 full/melody/off generation, audio-to-ABC covers, style or lyric changes, score editing, agentic reharmonization, melody preservation, singable lyric adaptation, and reproducible listening comparisons; also for YuE2 生成、翻唱、改编、改谱、换词和智能体编辑.

Оригинален текст · няма потвърден превод на български

Теми: ai · audio-generation · deep-learning · foundation-models · gpt · huggingface · llama · llms · music-generation · style-transfers · voice-cloning

Първо прочетете инструкцията.

Файлът се зарежда от посочения източник само по ваше действие. Можете да го прочетете тук, да го копирате или да го запазите като .md.

Оригинален SKILL.md
GitHub хранилище
Звезди на репото
9993
Лиценз
Apache-2.0
Последна проверка на записа
21.09.2026 г.

Добавете в своя инструмент.

Командата отваря избора на skills от репозитория. Прочетете файла и условията на автора преди изпълнение.

npx skills add multimodal-art-projection/YuE

Страницата копира текст. Не инсталира и не изпълнява нищо на устройството ви.

Откъс от SKILL.md

Началото на файла на автора, без превод.

YuE2 Music

Turn a musical request into a reproducible song and an audible comparison. Use released model interfaces. Retain an original song and its plan before making changes.

Recommended agent: GPT-6 Astra. YuE2 composes the default score and renders the audio; the agent operates the workflow and checks the results.

Instrumental generation and covers

For instrumental, no-vocals, or 纯音乐 requests, follow instrumental/SKILL.md. A simple style description is enough. Default to YuE2 planning, move every Vocal note to Ins, and render the revised score. Only write a new score with the agent when the user explicitly asks for agent composition. Given audio or ABC, use the instrumental cover workflow and retain the reference melody. Deliver playable audio and the actual prompt; the user does not need to specify voice names or inference parameters. Script and setup paths in that workflow are relative to instrumental/.

Choose the workflow

RequestWorkflow
Instrumental music from text, audio, or ABCInstrumental workflow → audio + prompt
Generate with editable melody and harmonyYuE2 cot="full" → ABC → song
Generate with a melody plan and free accompanimentYuE2 cot="melody" → chord-free ABC → song
Generate without symbolic planningYuE2 cot="off" → song; no editable ABC
Cover a recordingSheetSage2 → inspect/correct ABC → strip chords → YuE2 melody
Cover an ABC melodyInspect/convert native ABC → strip chords → YuE2 melody
Change harmony, instruments, tempo, structure, or lyricsCopy full plan → edit ABC/text → regenerate
Agentic editingExport plan/baseline → bounded editing agent → check invariants → render → compare
Analyze musical featuresUse MERT2 only when continuous features are needed
audio → SheetSage2 [loads MERT-v2-FullSong itself] → ABC
style + lyrics → YuE2 full/melody planning        → ABC
                                                  edit/validate
style + lyrics + ABC → YuE2 semantic generation → synthesis → latents → VAE → song
style + lyrics      → YuE2 off generation      → synthesis → latents → VAE → song

Do not feed public MERT feature tensors to YuE2 as codec tokens. YuE2 exposes no audio-reference, phoneme-alignment, or local-inpainting argument.

Set up the needed models

Целият SKILL.md в GitHub