Voices are synthesized by Puter — natural AWS Polly / neural / generative voices with no API key and no cost to you. Puter uses a user-pays model, so the first time you play a clip you may be asked to sign in to a free Puter account; normal personal use stays within its free allowance. Quality: Neural is a good default, Generative is the most lifelike (where a voice supports it), Standard is fastest.
Nothing you type or synthesize is stored on the server — no history, no library, no saved audio. Only your voice, language, quality, speed, and volume preferences are remembered. Downloads are generated in your browser from the current clip.