0.1.3

elyxlz/voxtral0.1.3Apr 7, 2026by zhu-han

AI Summary

This release focuses on compatibility improvements and bug fixes, including a relaxation of PyTorch version requirements. It introduces new options to the demo for better control and addresses issues with text processing and the torchaudio backend.

Key Highlights

Relaxed PyTorch version requirements for broader compatibility.
Added `no_asr` and `instruct` options to the demo.
Fixed text processing for non-verbal and special symbols.
Fixed torchaudio backend issues.

New Features

Added `no_asr` and `instruct` options to the demo.

Full Release Notes

## What's Changed
* Version 0.1.3 by @zhu-han in https://github.com/k2-fsa/OmniVoice/pull/61
  * relax pytorch version requirements 
  * Add tips in readme 
  * add no_asr and instruct options to demo 
  * fix text processing for non-verbal and special symbols 
  * fix torchaudio backend


**Full Changelog**: https://github.com/k2-fsa/OmniVoice/compare/0.1.2...0.1.3