AI voiceover & voice cloning
AI voiceover for every character in your script
abcVoiceover turns written scripts into finished voice audio. Clone a voice from a few seconds of recording, give every line its own speaker, and generate takes until each one sounds right, all in your browser.
Invite-only early access. Already in? Sign in.
What is abcVoiceover?
abcVoiceover is a web app for AI voiceover and voice cloning, built around scripts rather than single sentences. Instead of pasting one paragraph at a time into a text-to-speech box, you set up a project with a cast of voices, write or paste your script, and generate spoken audio line by line.
Each voice is cloned from a short recording, usually 5 to 15 seconds, with no model training and no waiting for a custom voice to be built. Every line remembers who reads it and which take you kept, so an audio drama with a dozen characters stays as organized as a two-line ad read.
From script to finished audio in four steps
-
Start a project
A project holds your scripts, your cast of voices and the people you're working with.
-
Clone your cast
Record in the browser or upload a WAV for each voice. 5 to 15 seconds of clear speech is plenty.
-
Paste your script
Start a line with a voice's name and a colon, like
Villain: Run!, and it's cast for you. -
Keep the best take
Generate, compare each new take with the last, and keep the best. Whole scripts finish in the background.
Built for scripts, not single sentences
Multi-voice scripts
Any line can be read by any voice in the cast, so one script can have as many speakers as it needs.
Voice cloning from short clips
Zero-shot cloning from one reference recording. Swap a clip any time; lines you've saved keep their sound.
23 languages
English, Spanish, French, German, Japanese, Hindi, Arabic and 16 more, chosen voice by voice.
Emotion and delivery controls
Start from Balanced, Expressive or Calm, then fine-tune emotion intensity, pacing and expressiveness.
Takes checked for you
Speech recognition listens back to every take, and anything that skips words or rambles is made again.
Sharing and clean exports
Share a project to view or edit, and download WAV takes in script order with a CSV of who says what.
Made for creators who work from a script
Audio dramas & podcasts
Cast a full ensemble and keep every character sounding the same from episode to episode.
Voicing audio fiction →Game dialogue
Voice NPCs, barks and quest lines, and re-generate them whenever the script changes.
Voicing game characters →Animation & YouTube
Character voices and narration for animatics, shorts and long-form videos, line by line.
Voicing videos →E-learning & explainers
Consistent narration for courses and training videos that's easy to update when content changes.
Narrating courses →Voice cloning, with consent built in
A voice belongs to a person. abcVoiceover only allows voices that are your own or that you have clear permission to use, and its terms rule out impersonation, scams and harassment. Your recordings and scripts are processed on a voice server the abcVoiceover team runs, aren't sent to an outside AI company, and aren't used to train models. How to use AI voices responsibly.
Common questions
How does voice cloning work in abcVoiceover?
It uses zero-shot voice cloning: the model listens to a short reference recording and speaks new text in that voice. Nothing is trained or fine-tuned, so a new voice is ready as soon as its clip is uploaded, and you can swap the clip at any time. The voice model is the open-source Chatterbox text-to-speech model, run on a server the abcVoiceover team operates.
How long should a voice sample be?
Aim for 5 to 15 seconds of one person speaking clearly, with no music or background noise. The minimum is 3 seconds and the maximum is 10 minutes, but longer isn't better: a clean, well-acted 10 seconds beats a noisy minute. Fast draft mode needs at least 5 seconds. Our guide on how to record a voice sample goes step by step.
Which languages does abcVoiceover support?
23: Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish and Turkish. You pick the language per voice. Fast draft mode is English only.
Is my voice data used to train AI?
No. Your recordings and scripts go only to the voice server the abcVoiceover team runs, to make your takes. They aren't sent to an outside AI company and aren't used to train models. The privacy policy has the details.
How do I get an invite?
abcVoiceover is invite-only during early access. Use the Request an invite button on the home page and tell us a little about what you're making, or ask someone who already uses it for a code. You type the code once, the first time you sign in.
Guides for better takes
Ready to hear your script?
abcVoiceover is in invite-only early access. Sign in with Google or Microsoft, then enter the invite code you were given.