Resona, a voice-rendering workbench

Field notes

Resona models a speaker from a 15–90-second recording, then lets the user write, render, listen, and revise from a single Studio screen. Its accounting is unusually explicit: renders consume credits by the second, while the homepage describes one credit as one minute of audio. Public model rows include English, Spanish, French, and Japanese voices, with each sample kept attached to the model it produced.