ACE-Step: A Step Towards Music Generation Foundation Model

Select previous generated input params

Audio Duration

-1 means random duration (30 ~ 240).

-1 240

Format

Check to enable Audio-to-Audio generation using a reference audio.

Enable Audio2Audio

Lora Name or Path

Reference Audio (for Audio2Audio)

Refer audio strength

0 1

Support tags, descriptions, and scene. Use commas to separate different tags.
Tags and lyrics examples are from AI music generation community.

Preset

Tags

Support lyric structure tags like [verse], [chorus], and [bridge] to separate different parts of the lyrics.
Use [instrumental] or [inst] to generate instrumental music. Not support genre structure tag in lyrics

Lyrics

Infer Steps

1 200

Guidance Scale

When guidance_scale_lyric > 1 and guidance_scale_text > 1, the guidance scale will not be applied.

0 30

Guidance Scale Text

Guidance scale for text condition. It can only apply to cfg. set guidance_scale_text=5.0, guidance_scale_lyric=1.5 for start

0 10

Guidance Scale Lyric

0 10

manual seeds (default None)

Seed for the generation

Scheduler Type

Scheduler type for the generation. euler is recommended. heun will take more time. pingpong use SDE

euler heun pingpong

CFG Type

CFG type for the generation. apg is recommended. cfg and cfg_star are almost the same.

cfg apg cfg_star

Use Entropy Rectifying Guidance for tag. It will multiple a temperature to the attention to make a weaker tag condition and make better diversity.

use ERG for tag

The same but apply to lyric encoder's attention.

use ERG for lyric

The same but apply to diffusion model's attention.

use ERG for diffusion

Granularity Scale

Granularity scale for the generation. Higher values can reduce artifacts

-100 100

Guidance Interval

Guidance interval for the generation. 0.5 means only apply guidance in the middle steps (0.25 * infer_steps to 0.75 * infer_steps)

0 1

Guidance Interval Decay

Guidance interval decay for the generation. Guidance scale will decay from guidance_scale to min_guidance_scale in the interval. 0.0 means no decay.

0 1

Min Guidance Scale

Min guidance scale for guidance interval decay's end scale

0 200

OSS Steps

Optimal Steps for the generation. But not test well

Text2Music Generated Audio 1

Text2Music Parameters

variance

0 1

retake seeds (default None)

Retake Generated Audio 1

Retake Parameters

variance

0 1

repaint seeds (default None)

Repaint Start Time

0 240

Repaint End Time

0 240

Repaint Source

text2music last_repaint upload

Upload Audio

Repaint Generated Audio 1

Repaint Parameters

Edit Tags

Edit Lyrics

edit seeds (default None)

Edit Type

only_lyrics will keep the whole song the same except lyrics difference. Make your diffrence smaller, e.g. one lyrc line change.
remix can change the song melody and genre

only_lyrics remix

edit_n_min

0 1

edit_n_max

0 1

Edit Source

text2music last_edit upload

Upload Audio

Edit Generated Audio 1

Edit Parameters

extend seeds (default None)

Left Extend Length

0 240

Right Extend Length

0 240

Extend Source

text2music last_extend upload

Upload Audio

Extend Generated Audio 1

Extend Parameters

·

Built with Gradio logo

·