Skip to content

TangoFlux

Generate 30-second audio from text in seconds

Visit official site

What TangoFlux is

The model is described as an efficient text-to-audio system with 515M parameters, using a preference optimisation framework to improve alignment. Code and models are open source.

What you can do with it

  • Generate up to 30 seconds of 44.1kHz audio
  • Produce stereo background music and effects
  • Apply iterative preference data for alignment

Getting started

  • Download the open-source code from GitHub
  • Install the dependencies and environment
  • Run the code with a text input to produce audio

Pros & cons

✓ What we like

  • Iterative preference data generation for alignment
  • Handles audio tasks up to 30 seconds

! What to watch out for

  • The listed run uses a single A40 GPU
  • Alignment work runs through the CRPO framework

FAQ

How fast is generation?

The material states 3.7 seconds for up to 30 seconds of 44.1kHz audio on one A40 GPU.

Is the code available?

The material states all code and models are open source.

Last reviewed: 2026-11-15

More AI audio tools tools

View all →

How we review