Generate an AI voiceover from a script in DepCut

Write the lines, choose a voice, press a button. The Voice tab turns a script into narration and places it on your timeline. Voiceovers run on your DepCut account and each generation spends credits.

Open the Voice tab and pick a model

In the left panel, open AI Gen, choose Audio, and stay on Voice. The panel starts with the Audio model menu, because the model decides which voice controls appear below it. The menu lists five models:

• DepCut Voices is fast hosted speech with a built-in cast of voices and delivery directions.
• Eleven v4 and Eleven v3 are expressive models that support audio tags and a wide emotional range.
• Eleven Multilingual v2 targets voiceovers in many languages.
• Eleven Flash v2.5 is a low-latency model for quick turnarounds.

The panel opens on Eleven v3. Switch models any time and generate again to compare.

Choose a voice

With an Eleven model, the Voice menu lists the available voices. Type in Search voice to find one, use the chips to filter by All, Recent, Male or Female, and hover a voice to hear a sample. Press + to look up a voice by its ID or name and add it.

With DepCut Voices, the Speaker voice menu opens a grid of portraits. Hover a portrait to hear that voice, and click to pick it. Each voice carries a one-word character, such as Sofia (Warm), Charlie (Informative) or Gus (Gravelly). The play button beside the menu samples the voice you have selected.

Direct the delivery

DepCut Voices has a Voice direction box. Describe how the lines should sound: "Say warmly, like an old friend." The preset menu next to the box offers Warm, Energetic, Documentary, Movie trailer, News anchor, Whisper and Bedtime story, plus a Custom option that clears the box for your own wording.

You can also ask for another language in the direction. DepCut translates the script and the voice speaks it in that language. Eleven models read the script exactly as you wrote it, in whatever language it is written in.

With a direction written, the sample button generates a fresh sample so you can hear the delivery. That sample spends credits. Without a direction, it plays a prebuilt clip.

Write the script and generate

1. Click into Script and write what the voice should say. Press the microphone button to dictate it.
2. Press Generate audio.
3. Wait for the status line to finish. Several generations can run at once.

The voice appears under Generated audio, labeled with the line it speaks, and a clip is added to the soundtrack at the playhead. Set the playhead before you press the button.

New voiceovers arrive with Duck others set to 40 percent, so music and other sound drop to that level while the voice speaks. Select the clip and open the Duck tab to change it, or open Volume to set the level of the voice itself.

Write for the ear

A script that reads well on a page can feel stiff when spoken.

• Use short sentences. One idea per sentence is easy to follow when it is heard.
• Read the script aloud before you generate. Wherever you stumble, rewrite the line.
• Punctuate the way you would speak. Commas and full stops give the voice places to pause.
• Generate in sections. Short passages are quicker to redo, and you can place each one where it fits the picture.

Next step

Open DepCut, choose a voice, and generate the first line of your narration. Listen to it over the picture before you commit to the whole script.