Your voice, from a few seconds of audio.

A voice is a timbre, not a recording.

Voice cloning makes a synthetic voice that sounds like a particular person. Older systems trained a model on hours of one speaker. Versa is zero-shot: it listens to a short reference clip and speaks any text in that voice, with nothing to train.

Because the voice is a timbre rather than a language, a clone recorded in Spanish also speaks English, Catalan or German, so one brand voice can serve every market.

• From a clip3 to 30 seconds, recorded in the dashboard or uploaded. A quiet room gives the cleanest clone.
• EverywhereUse its clone_… id anywhere a voice is accepted: the speech API, streaming, and voice agents.
• Yours aloneA cloned voice is visible and usable only from the account that made it.
• PriceNothing extra. The clone is billed like any voice: $4.10 per million characters, or inside an agent’s minute.

Then use it like any voice.

curl -X POST "https://api.lokutor.com/tts/synthesize" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text":"Hola, soy la nueva voz de tu marca.","voice":"clone_YOUR_VOICE_ID","language":"es"}' \
  --output speech.wav

import requests

r = requests.post(
    "https://api.lokutor.com/tts/synthesize",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    json={"text": "Hola, soy la nueva voz de tu marca.", "voice": "clone_YOUR_VOICE_ID", "language": "es"},
)
open("speech.wav", "wb").write(r.content)

Cloned voices unlock with an account’s first top-up. Only clone a voice you own or have permission to use.