Give a character a consistent voice

tendre.red can make a character's spoken clips consistently use one voice. It works from a short reference sample, not by typing a script into a text-to-speech box, and the conversion currently runs in the cloud.

1. Import a sample

Open Library > Voices and click Import a voice. Pick a clean recording, 10 to 30 seconds, one voice, minimal background noise. The file is copied into the app's voice library. Do not import a real person's voice without their consent.

2. Manage the library

Rename an entry by double clicking its name, preview it with the Listen button, and see which characters currently use it in the Used by column. Delete a voice and any character linked to it falls back to the engine's default voice.

3. Assign it

Attach the voice to a character's profile so their spoken videos adopt that timbre by default, or pick it directly from the Voice dropdown when generating a clip.

4. Generate with sound

In the Video tool, choose the Cloud engine, set the model to LTX 2.3 Full, and turn Sound on. The faster LTX 2.3 Turbo model also renders with sound but always uses the engine's default voice, the Voice dropdown only shows up on Full. tendre.red converts the generated dialogue to the selected voice's timbre and lip syncs it to the character on screen. Local, GPU only, rendering also produces native sound through LTX 2.3, but without this voice conversion step.

To reuse a voice without re-importing the sample every time, see Sound LoRA. To add ambience or effects around the dialogue, see Sound effects.