All docs

Narration and voice

Narration and voice

Choose the language and voice when you create a video, then edit, regenerate or record the narration for each step in the editor.

Last updated

You pick the language and voice when you create a video, and the build records the narration for you. After that, the editor's Narration panel lets you rewrite any line, generate its voice again, change the voice for the whole video, or record lines yourself.

When you create a video

The create screen has these settings:

  • Voiceover: Narrated (the default) means "The build speaks every scene's script, so the finished video arrives already voiced." None means "Scenes are captioned but silent."
  • Language: Auto (the default) follows the language of your topic or pasted material. Or pick English, Spanish, Portuguese, French, German, Italian, Greek, Japanese, Chinese, Korean, Hindi or Arabic, and every title and line is written in that language, whatever language your material is in.
  • Voice: 20 built-in voices. Alloy is the default. A few show first, and + N more shows the rest. Click ▸ Preview voice to hear one.
  • Sound effects: None (the default), Soft, Playful, Retro 8-bit or Mechanical. Short motion cues play as shapes appear and connect. Click ▸ Preview sound to hear them.

The voices are Alloy, Ash, Ballad, Coral, Echo, Fable (British), Nova, Onyx, Sage, Shimmer, Verse, Sarah, Liam, Lily, Aria, Roger, George (British), Alice (British), Charlotte and Callum.

Cartoon skits work differently: each character has their own voice, and the skit speaks the language your text is in. See Edit a cartoon skit.

The Narration panel

In the editor, open Narration on the right rail. Narration is set per step: each step of a scene has its own line in the "Narration for this step..." box. See The timeline for what a step is.

Change a line

  1. Select the step and edit its text.
  2. Click Generate voice from text. The new line is not spoken until you do this.
  3. If the new line is longer or shorter, click Fit Steps to Narration (see below).

To voice several lines at once, use Generate All for every line in the scene you are on, or Generate All Scenes for the whole video. Clear all narration removes the narration from the current scene.

Change the voice for the whole video

  1. In the Narration panel, pick a voice. Click Hear this voice to try it, or Browse accented voices to search by accent and gender.
  2. Click Generate All Scenes to record every line again in the new voice. Picking a voice on its own does not change lines that are already recorded.

Record your own voice

  1. Select a step and click Record your own voice. Allow microphone access when your browser asks.
  2. Read the line, then click Stop recording. Press Play to hear it.

The recording is used for that step only, so you can mix your own voice with generated lines. To go back, use Replace recording with synthesized voice. The step's More menu also has Enhance (remove noise), Studio voice and Clear audio.

The Clean up recordings (denoise, gate, de-ess, loudness) switch in the panel is on by default, so each new recording is cleaned when you stop recording. With it on you can pick a Tone (Natural, Podcast, Warm or Bright) and turn on Max denoise for stronger noise removal. Turn the switch off before recording if you want your raw audio. A quiet room still gives the best result.

If you see "Microphone permission denied.", allow the microphone for this site in your browser settings. "Recording is not supported in this browser." means you need a different browser.

Keep timing in step with narration

Steps do not retime themselves when narration changes. Fit Steps to Narration sets each step's length to its generated narration plus a 0.2 s buffer, never shorter than the step's own animations. Run it after you generate new narration.

A silent scene

A scene with no narration shows a 🎤 Narrate this scene button that takes you to the Narration panel.

Related


All Rights Reserved.