lucataco / zeta-editing

Zero-Shot Text-Based Audio Editing Using DDPM Inversion

  • Public
  • 1.8K runs
  • L40S
  • GitHub
  • Paper
  • License
Iterate in playground

Input

Video Player is loading.
Current Time 00:00:000
Duration 00:00:000
Loaded: 0%
Stream Type LIVE
Remaining Time 00:00:000
 
1x
*file

Input Audio File

string
Shift + Return to add a new line

Describe your desired edited output

Default: "A recording of an arcade game soundtrack"

integer
(minimum: 15, maximum: 85)

Lower % returns closer to the original audio, higher returns stronger edit

Default: 45

string

Choose the audio version to return

Default: "cvssp/audioldm2-music"

string
Shift + Return to add a new line

Optional: describe the original audio input

Default: ""

integer

Number of diffusion steps, higher values(200) yield high-quality generations

Default: 50

number

Source Guidance Scale

Default: 3

number

Target Guidance Scale

Default: 12

integer

Random seed

Output

Video Player is loading.
Current Time 00:00:000
Duration 00:00:000
Loaded: 0%
Stream Type LIVE
Remaining Time 00:00:000
 
1x
Generated in

Run time and cost

This model costs approximately $0.11 to run on Replicate, or 9 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia L40S GPU hardware. Predictions typically complete within 108 seconds. The predict time for this model varies significantly based on the inputs.

Readme

Zero-Shot Unsupervised and Text-Based Audio Editing Using DDPM Inversion

Technion - Israel Institute of Technology

img

Abstract

Editing signals using large pre-trained models, in a zero-shot manner, has recently seen rapid advancements in the image domain. However, this wave has yet to reach the audio domain. In this paper, we explore two zero-shot editing techniques for audio signals, which use DDPM inversion on pre-trained diffusion models. The first, adopted from the image domain, allows text-based editing. The second, is a novel approach for discovering semantically meaningful editing directions without supervision. When applied to music signals, this method exposes a range of musically interesting modifications, from controlling the participation of specific instruments to improvisations on the melody.

Note: For now use input audio wav files