v0.2.3

cjpais/Handyv0.2.3Jul 3, 2025by github-actions[bot]

AI Summary

This release changes the model quantization strategy to improve performance and fixes a keycode bug. It is a breaking change for users of medium or turbo models who need to redownload them.

Key Highlights

Changed model quantization for Medium and Turbo models
Medium model switched to q4_1 quantization
Turbo model switched to unquantized
Fixed keycode bug

Breaking Changes

Users of Medium or Turbo models must redownload models due to quantization changes.

New Features

Model quantization changes
Keycode bug fix

Full Release Notes

This is a small breaking release in the case you were using the medium or turbo models. You will need to redownload them.
I switched to different quantizations of the models to help improve transcription performance for each. For whisper medium we are using q4_1 and for whisper turbo it is unquantized.

## What's Changed
* Fix the keycode bug by @cjpais in https://github.com/cjpais/Handy/pull/31
* Different default models are also used for Medium and Turbo.

**Full Changelog**: https://github.com/cjpais/Handy/compare/v0.2.2...v0.2.3