AI voice changer that keeps your delivery
Say the line your way. Our AI voice changer re-speaks it in one of your HeyGen voices, with the same words, pacing and pauses.
Conversational
Everyday dialogue, re-spoken in a new voice.
Original voice
Brody · laptop mic
Flip
mid-sentence
Changed voice
Elliot · HeyGen Voice
“Honestly, the best part of the trip was the food. We ate our way through the whole city.”
Change the voice, not the delivery
Your words and delivery, exactly
It starts from your recording, not a script. Your emotion, emphasis and cadence come through in the new voice.
Your timing
The output is the same length as your recording, so it lines up with your existing edit.
Background noise removal
Recorded on a laptop or phone? Switch it on to clean up the take before it’s converted.
Use the voice changer in avatar videos
In Studio, the voice changer is called Voice Mirroring. Record your take for a scene, and the avatar lip-syncs to it in the scene’s voice.
- 1Click Upload audio in the script area.
- 2Record or upload your performance.
- 3Turn on Voice Mirroring and add the audio.
What you can do with a voice changer
Fix a voiceover line without a re-session
Re-record one sentence on any mic and convert it to your clone, so it matches the rest of the take.
Many people, one brand voice
Anyone on the team can record the script. Every voiceover comes out in the same narrator voice.
Direct the performance yourself
Act out the timing and emphasis, then hear it in the character’s voice.
Consistent course voiceover
Experts record lessons in their own words. Every module ships in one consistent voice.
How to change your voice with AI
- 01
Record or upload
Record in the browser or upload MP3 or WAV. The API also takes FLAC, OGG, M4A and AAC, up to 5 minutes.
- 02
Pick a voice
Choose a voice from your workspace, or clone your own from one recording. It’s ready in minutes.
- 03
Use it in a video or your app
Drop it into a Studio scene, or call the API and get back a WAV the same length as your take.
Voice changer API
- Synchronous, so there’s nothing to poll.
- Send audio as an asset, a URL or base64.
- Safe retries with idempotency keys.
Synchronous, so there’s nothing to poll.curl -X POST "https://api.heygen.com/v3/models/audio/sts" \ -H "X-Api-Key: $HEYGEN_API_KEY" \ -H "Content-Type: application/json" \ -H "Idempotency-Key: take-01" \ -d '{ "voice_id": "<your_instant_voice_id>", "audio": { "type": "url", "url": "https://example.com/take.mp3" }, "remove_background_noise": true }'
Pay per second of output audio
Instant voices are free to create during the preview.
$0.12per minute
AI voice changer FAQ
What is an AI voice changer?
It’s a tool that takes a recording of someone speaking and re-speaks it in a different voice. HeyGen’s voice changer uses speech to speech, so your words, pacing and pauses carry over and only the voice changes.
Where can I use HeyGen’s voice changer?
In two places. In HeyGen Studio it’s called Voice Mirroring: record or upload audio for a scene, and the avatar lip-syncs to it in the scene’s voice. Developers can use the Speech to Speech API to turn a recording into a standalone WAV file.
What can I use an AI voice changer for?
Voiceovers, podcasts, ads, courses and character work. You perform the line once, the way you want it, and hear it in the voice the project needs. In Studio, it can go straight into an avatar video.
How do I change my voice in a HeyGen video?
Open your video in Studio, select the scene and click Upload audio in the script area. Record or upload your take, turn on Voice Mirroring in the Confirm Audio dialog, then click Add audio. The scene uses the voice set in the Avatar & Voice panel, and HeyGen transcribes your recording so you can check the words.
Can I change the voice on a scene that already has a script?
Yes. Open the scene’s ⋯ menu and choose Mirror voice. Your script shows up as a teleprompter, so you can record or upload a performance for that scene.
Which voices can I convert into?
In Studio, the voice selected for the scene. Through the API, any active Instant Voice your workspace owns. Not every voice in your library is an Instant Voice, and the API’s voice list shows which ones you can use. Professional voice clones aren’t supported by the API yet.
What files can I use?
In Studio, record in your browser or upload an MP3 or WAV file up to 100 MB. Through the API, send WAV, MP3, FLAC, OGG, M4A or AAC up to 5 minutes long: up to 100 MB as a HeyGen asset, 32 MB by public HTTPS URL or 16 MB as base64.
What do I get back?
In Studio, the converted audio goes straight into the scene and the avatar lip-syncs to it. Through the API, you get a link to a WAV file the same length as your recording, with the same words and timing, in the voice you picked. The API returns audio only, not video.
Does the voice changer rewrite or translate what I say?
No. It keeps your words and timing exactly as you recorded them. To translate a video into another language, use HeyGen Video Translation.
How is a voice changer different from voice cloning?
Voice cloning creates a reusable voice. The voice changer uses your recording to drive the words, timing and delivery of a voice you already have, including your own instant clone.
How is speech to speech different from text to speech?
Text to speech generates audio from a script. Speech to speech starts from a real performance, so your timing and delivery carry into the new voice.
Does it work in real time?
No. It converts a finished recording, not live audio. Through the API, a conversion can take up to 150 seconds, so set your client timeout to at least 160.
How do I get the best results?
Record one speaker, with no music or overlapping voices. Make your pacing, emphasis and pauses intentional, because they carry into the new voice. Start with short segments, and preview the scene before you generate the full video. On the API, turn on background noise removal for noisy takes.
How much does it cost?
On the API, $0.12 per minute of output audio, billed per second.
