Make a voiceover

Voiceovers are made on Voice. Your takes are under My voiceovers, and the Templates tab has scripts to start from, each with a voice of the kind it names and a delivery to match: a template fills in the bar, and nothing runs until you press Generate.

Make a take

1. Write the script in the bar at the foot of the page. Under it, a line estimates how long it runs aloud and counts its characters against the longest one take reads.
2. Choose a voice. Open the Voice chip, search the voices by name or trait, filter them by Male or Female, and play a sample before you pick. "Top" marks the featured voices.
3. Pick a model: Eleven v4, marked Best, Eleven v4 Turbo, or Eleven v3. Each row shows what your script would cost on it. See How long a voiceover can be, and what each voice model does.
4. Set the language, if you need to. Auto-detect works it out from the script's words; pick a language and the take is read in that one.
5. Pick a format: MP3 at 128 kbps, the standard, or PCM at 16, 22 or 24 kHz, which downloads as WAV.
6. Adjust the delivery, if you like, under Settings: Stability, from Expressive to Steady, and Clarity, from Loose to Close to the voice. Natural speech sits around stability 0.3 to 0.5 and clarity 0.7 to 0.9. Reset puts them back.
7. Press Generate, or ⌘ Return (Ctrl Enter on Windows) in the script.

The take's card appears at the top of My voiceovers straight away. When it's ready it's loaded into the player at the foot of the page, ready to play; a take never starts playing on its own. The script clears for the next one, and you can press Generate again while a take is being made.

Direct the delivery

Directions in square brackets are performed, not read aloud: [whispers], [laughs], [excited], [pause]. All three models perform them.

• Tags opens a list of them by kind: Emotion, Reactions, Delivery, Character and Sound effects. Pick one to put it in the script where the cursor is, or type your own: any short direction in brackets works, such as [leaning in].
• Auto-tag adds tags where a voice actor would change the delivery, and leaves your words as they are. It's free. Undo puts the script back, until you edit it.
• CAPITALS add emphasis, and an ellipsis… adds a pause.

Two or more voices

Press Add speaker under the script. The script becomes the first line, read by the voice you picked, and a new line follows for a second voice. Each line has its own voice chip, and × takes a line out; down to one line, the bar goes back to a single voice. A dialogue is made as one take, in one file, with up to 10 voices and 100 lines.

When you press Generate on a dialogue, its voices stay in the bar for the next exchange, and only the words clear.

What it costs

A take is priced by the length of its script: every started 100 characters at the model's rate. A thousand characters cost 8 credits on Eleven v4 and 4 credits on Eleven v4 Turbo. Audio tags count as characters too. A dialogue is priced the same way, by all its lines' characters.

The price is beside Generate before you press, and you're never charged more than that. A take that fails gives its credits back automatically, and its card says how many.

Your takes

Press a take's card to play it, or press or drag along its track to start part of the way in. Each card has Download, and under ⋯, Use this script, which puts its words back in the bar with its voice, or every line with its voice for a dialogue, Copy link and Delete. The player at the foot steps between takes and downloads the one playing.

My voiceovers shows the takes made on this page. Voiceovers made on a board are in your Library.

To use a voiceover in a video, see Put a voiceover or a song in a video.