Skip to content

YuE

Turn lyrics into a full song with vocals and accompaniment

Visit official site

What YuE is

The model is described as an open-source music generation model that turns given lyrics into a full song of up to five minutes, including vocals and accompaniment. It is built by a university and a research team and uses techniques such as a semantically enhanced audio tokenizer, a dual-token technique and lyric chain-of-thought.

What you can do with it

  • Generate a full song of up to five minutes from lyrics
  • Produce vocals and accompaniment in one generation
  • Work with English, Chinese, Japanese and Korean lyrics
  • Choose among styles such as pop, rock, jazz and rap

Getting started

  • Download the model checkpoints from the GitHub page
  • Prepare lyrics text and pick a language and music style
  • Split the lyrics into several segments with the provided tools
  • Feed the lyric segments into the model to generate music segments

Example uses

  • Generate a Chinese pop song whose lyrics come from GPT
  • Produce an English metal song named Step Back
  • Produce an English jazz song named Quiet Evening

Pros & cons

✓ What we like

  • The model is open source and currently free
  • Multiple model checkpoints and upsamplers are provided

! What to watch out for

  • Song generation starts from prepared lyrics
  • Lyrics are split into segments before being fed to the model

FAQ

How long can a generated song be?

The material states the model generates full songs of up to five minutes, including vocals and accompaniment.

Which languages are supported?

English, Chinese, Japanese and Korean lyrics are listed.

Last reviewed: 2026-11-15

More AI music creation tools

View all →

How we review