Skip to content

OmniAudio-2.6B

A 2.6B audio language model for edge devices

Visit official site

What omniaudio-2-6b is

The model is described as a compact audio language model that unifies speech recognition and language processing for edge devices.

What you can do with it

  • Answer a spoken question about a task
  • Hold a voice conversation about your day
  • Write a short creative piece from a spoken prompt

What to watch out for

  • The Nexa SDK is installed before running
  • A Streamlit interface can be started with a flag
  • The q4_K_M build needs 1.30GB RAM and 1.60GB storage

Pros & cons

✓ What we like

  • Voice Q&A, dialogue and creative content generation are listed
  • Performance on consumer hardware is reported as 5.5 to 10.3 times

! What to watch out for

  • The Nexa SDK needs to be installed
  • The q4_K_M build needs 1.30GB RAM and 1.60GB storage

FAQ

What inputs are supported?

The material states text and audio inputs are handled together.

How is it run?

The card lists a nexa run omniaudio command in the terminal.

Last reviewed: 2026-11-15

More AI audio tools tools

View all →

How we review