Design Tai's voice from your own description — generate, tweak, save, and switch between voices you've made.
Generate 3 original deep, gravelly, American candidates and pick one.
Auto matches what you speak. Pick a language to pin replies, or say “habla español” / “speak Spanish” any time — sticky until you switch again.
Companion library first: warm, sultry, assertive, calm, playful, professional, soft. Tap Preview to hear a sample, Select to pin. Next spoken reply uses it. Radiant is the recommended default.
Companion voices
Warm mid-tone. Rich. Feminine.
Warm mid-range feminine companion — rich, unhurried, womanly without breathiness. Moderate stability keeps it natural and close.
Deep. Warm. Sultry.
Low, husky female voice with a warm sensual edge. Raised stability and similarity keep it smooth and unhurried — the deepest female timbre in the set.
Assertive. Confident. Direct.
Clear, confident American female delivery — firm pace, decisive consonants, zero hesitation. Built for decisive I-have-this energy without shouting.
Calm. Steady. Reassuring.
Even, composed American female voice — measured pacing, soft edges, easy to listen to for long answers. Low drama, high clarity.
Playful. Bright. Upbeat.
Bright, lively feminine register with quick lift on key words — light, friendly, a little mischievous without going cartoon.
Professional. Clear. Composed.
Polished professional female voice — crisp diction, neutral American accent, boardroom-calm confidence without coldness.
Soft. Intimate. Smooth.
Soft, velvety feminine tone — gentle volume shape, intimate pacing, smooth vowels. Quiet confidence rather than projection.
Alternates
Deep. Stoic. Cinematic.
An original deep neural voice tuned for gravitas and control. High stability keeps delivery steady and unhurried; elevated style adds cinematic weight without theatrics.
Controlled. Authoritative. American.
Deep US male with a measured, no-nonsense delivery. Maximum stability, zero style uplift — every word lands flat and final.
Warm gravitas. Steady.
A warm, grounded American baritone. Still authoritative, but with lived-in depth — less edge, more weight.
Private. No cloud. Always works.
Your device's built-in voice engine. Zero network, zero latency. Lower quality but fully private — sensitive content stays on-device regardless of your cloud voice setting.
Selection is applied on the next spoken reply (ElevenLabs TTS). Preview plays a short sample. On-device is always the fallback when cloud is unavailable. For a fully custom original design, use Voice Studio above.
The voice on the live call. Takes effect on your next call; saved to your account, so every device — including the TV — uses it.