Voices and Voice Lab

Choose a stock voice, clone your own from a short recording and keep every voice under your control.

Updated

Stock voices

The library has voices for every tone: calm, upbeat, formal, warm. Each one comes in several languages and can be adjusted for speed, pitch and expressiveness.

Cloning a voice in Voice Lab

  1. Open Voice Lab and choose New voice.
  2. Upload a recording or record in the browser. Speak naturally, in a quiet room.
  3. The voice owner reads the consent sentence shown on screen.
  4. Give the voice a name. It is ready in about a minute.

Getting a good recording

  • Use one speaker, no music and no background noise
  • Speak the way you want the voice to sound in your product
  • Avoid heavy compression and phone call audio

Watermarking

Everything generated with a cloned voice carries an inaudible watermark, so audio can always be traced back to the workspace that made it.

Questions about voices and voice lab

How long does a recording need to be?
Thirty seconds of clean speech is enough to start. Two minutes gets you noticeably closer to the original.
Can I clone someone else's voice?
Only with their spoken consent. The voice owner reads a random sentence aloud, and we match it to the sample before cloning.
Who can use a cloned voice?
Only your workspace. Cloned voices cannot be shared, exported or used from another account.
Can I delete a voice?
Yes, at any time from Voice Lab. Deleting a voice also deletes the samples it was trained on.