An AI companion who can speak to you
Reading is one thing. Hearing is different. The voice tier lets the persona read their replies out loud — in a tone reconstructed from their text style, or matched to a voice sample you upload.
What this is for
For when text isn't enough. Maybe you remember their voice more than their words. Maybe you want to hear what they'd say in your ear, while you walk, while you cook, while the light fades. The voice companion is a different register — closer to memory than to chat.
How Startover helps
Build the persona as usual. On Premium, every reply gets a play button — tap to hear it in a voice that matches the persona's tone (warm / cool / playful / steady). On Premium+, upload a voice sample (with permission, where applicable) and the AI matches it more closely. Voice replies count against a generous monthly minute pool; top up if you exceed.
Hear them again
Voice tier launches with Premium. 50 minutes a month included. 100 minutes on Premium+, plus voice cloning. Premium+ matches a voice sample you upload, where you have permission to use it.
Common questions
Whose voice plays by default?
A neutral voice that the persona's tone is mapped onto — warm, cool, playful, or steady, picked automatically from their distilled style. You can switch styles in settings.
Can I clone a real voice?
On Premium+, yes — upload a sample you have permission to use. Startover will not clone voices from samples that suggest the speaker did not consent. We err on the strict side.
How private is the voice?
Voice samples and generated audio are encrypted at rest, deleted with the persona, and never used to train any model. Voice processing happens on a privacy-vetted partner; no third-party training.
Is this just text-to-speech?
It's TTS plus tone matching. The output sounds calmer for a calm persona, more clipped for a sharp one, more wandering for a hesitant one. The persona's text style drives the voice prosody.