Originally released in 2020, but revised with some quality of life improvements:
Stabilised formant process application across CPU architectures
Use auto makeup gain in favor of compression algorithm to keep in and out volume in check
When (re)syncing vowels, oscillator positions are now realigned
Added dry/wet mix control
Added input threshold control
Added bypass control
Added linear smoothing to vowel changes
Added better support for textual input of parameter values outside of the UI
LFO's are reset and synced on host sequencer start
As before, Transformant remains free to use and fully open source. Windows and macOS versions can be downloaded from the webpage.