Voice Swap
Keep your delivery, change the voice. Upload a recording and a sample of the voice you want — every word, pause and inflection stays exactly as you performed it.
Keep your delivery. Change the voice.
Record the line the way you want it heard — the pacing, the emphasis, the emotion — then swap in a different voice. Every word, pause, and inflection stays exactly where you put it. Only the voice changes.
This matters because most voice tools throw your performance away. Type a script into a text-to-speech tool and you get the machine’s reading of it, not yours. Here you act the line, and the delivery survives the swap.
How to swap a voice
- Upload your recording. Speech, audio only (mp3, wav, flac, m4a, ogg), up to 5 minutes. Split longer sessions first.
- Upload a voice sample. Ten seconds to two minutes of the target voice, one speaker, no music behind it. About a minute gives the best match.
- Run it. Most clips finish in about a minute.
- Download your track. One audio file in the new voice.
What you can do with it
- Re-voice a scratch track you recorded yourself into the voice the final cut needs
- Give a character a voice by acting the line, then converting it into a sample of that character
- Match a voiceover to a brand voice from one clean sample, without booking a session
- Audition the same line in several voices by swapping the sample
Your performance is the point
Pace, emphasis, hesitation, the breath before the punchline — all of it comes from your take, not from the sample. The sample supplies timbre and nothing else.
So act the line properly. A flat read converted into a great voice is still a flat read, and a well-acted take converted into an ordinary voice usually sounds better than the reverse. If you have ever recorded a scratch vocal just to get the timing right, this is that, for voice.
For the full technical detail on how the conversion works, see the Voice Changer model page.
Tips
- The sample decides the result. A clean, dry, single-speaker clip gives a faithful voice; a noisy or mixed one gives a muddy result.
- Record the sample directly. A clip lifted from a published video may be refused.
- Clean up noisy input first. Run either file through the Voice Isolator if there is noise behind it, or switch on “Remove background noise” in the settings.
- Use voices you have the right to use. Your own, a collaborator’s with their consent, or a licensed one. Every cloned voice is screened and some samples are refused.
FAQ
What is voice swapping? Also called speech-to-speech conversion. You supply a recording and a target voice, and the recording comes back in that voice with the original performance intact. It is different from text-to-speech, which generates a reading from written words rather than converting one you already made.
How is this different from Voice Cloner? Voice Cloner turns typed text into speech in a cloned voice. Voice Swap starts from audio you already recorded. If you want control over delivery, record it and swap the voice; if you just need a script read aloud, use Voice Cloner.
Do I need a long sample? No. Ten seconds works, and about a minute is the sweet spot. Past two minutes there is no further gain.
Can I change my voice in real time? No. This works on files you upload, not on a live microphone, so it is built for recorded audio — voiceover, dialogue, narration — rather than live calls or streaming.
Can I use it on singing? Use the AI Song Cover Generator instead. It tracks pitch and melody, which speech conversion does not.
Is it free to try? Yes. Runs cost credits, and new accounts start with some.
Whose voice am I allowed to use? Your own, a collaborator’s with their permission, or a voice you have licensed. Samples are screened and some are refused; if that happens, record a fresh sample directly rather than re-uploading the same clip.
What audio formats does it take? mp3, wav, flac, m4a, and ogg, up to five minutes per run. Longer recordings need splitting first.