Voice Studio#

Voice Studio turns a text into a voiceover video: you type the text, pick an AI voice and a look, and get a video with a moving background, bars that move with the voice, and optional subtitles and music. On paid plans you can also download the audio track on its own, as an MP3.

No API key, JSON or template is needed.

It lives at json2video.com/dashboard/voice-studio, under Voice Studio in the dashboard menu.

Looking for a voice? Browse the AI voice gallery, then use the voice you like in Voice Studio.

Creating a voiceover#

Voice Studio opens on the list of your voiceovers. Click + New voiceover to open the editor. If you have no voiceovers yet, the editor opens directly.

1. Write the script. The text goes in a speaker block. Each block has one voice and its text. + Add speaker adds another block, with its own voice; the blocks are spoken one after another.
2. Pick the voice. Click the voice to open the voice window. Search by name, or filter by provider (ElevenLabs or Azure), language and gender. The play button plays the voice's own sample.
3. Choose the look. Three backgrounds, each with its audiogram: Blue green, Pastel and Light lines. Turn Subtitles on to show the words on screen, in one of three styles. Pick the shape: Vertical (1080×1920), Square (1080×1080) or Wide (1920×1080).
4. Add music, if you want it: No music, Quiet, behind the voice or Medium. With music on, pick a track from the library; you can listen to each track first.
5. Click Create voiceover.

The voiceover renders like any other video. It usually takes about as long as the voiceover itself, often under a minute. You can leave the page: the list shows the voiceover as Creating… and then as ready.

When it is ready, the video plays in the page, with three buttons: Download video, Open video and Download audio track.

Your own text is not spoken before you create the voiceover. Only the voice's sample can be played beforehand.

Expression: tags in the text#

To change how a voice speaks, write a tag in square brackets inside the text:

[excited] We are live! [short pause] [whispers] But keep it a secret.

The Add expression button under each block lists what that block's voice can do, and inserts the tag for you.

A tag the voice cannot do is shown greyed out and is left out. It is never read aloud, and it never makes the voiceover fail.

ElevenLabs voices do not always follow a tag. If one is ignored, try another wording, add punctuation, or try another voice.

Pick a voice that speaks the language of your text. ElevenLabs voices read many languages. Most Azure voices read one: if the text is in another alphabet, the editor warns you, because the voice may not be able to read it.

Plans, limits and cost#

A voiceover costs 1 credit per second, like any video. The voices cost no extra credits. See Billing for plans and credits.

While you type, the editor estimates the length of the voiceover from the number of characters (about 15 per second, fewer for languages such as Arabic, Hebrew, Chinese, Japanese or Korean; tags are not counted). If the estimate is over what your plan allows, the editor tells you before anything is created, so no credits are spent on a voiceover that would be refused.

Videos and audio files are kept for 7 days, like every render. After that, the voiceover stays in the list with Edit, so you can create it again.

Using your own ElevenLabs key#

If your account has its own ElevenLabs key set as the default in Connections, the editor shows Use my ElevenLabs key, switched on. ElevenLabs then bills your own account, and you can paste the ID of any voice of your ElevenLabs account in the voice window.

Without your own key, nothing about keys is shown and JSON2Video's key is used.

Common questions#

I only want the audio. Voice Studio always creates a video. The MP3 of its audio track is a download for paid plans. On the free plan you can download the video, with the watermark.

Download video opened a new tab. For very long videos the page opens the video instead of downloading it. Use your browser's Save video as… to save it.

Where else do my voiceovers appear? They are normal videos, so they are also listed in Render logs and returned by GET /movies.

The same result from the API#

There is no audio-only endpoint. A voiceover is a movie with voice elements, an audiogram element and "audio-track": true at the movie level. On paid plans, the MP3 URL comes back as audio_url in the response of GET /movies.