All products

Change a recorded voice with AI in ElevenLabs

Use an authorised recording and target voice to audition a performance with a different vocal sound while retaining the delivery of the source.

To change a recorded voice with AI in ElevenLabs, give voice-changer a performance you are authorised to process and a target voice available to your account. A character designer might record a line themselves and try it with a permitted stock voice. The source provides the timing, expression and pauses that make the line work.

Listen to the original recording first. If the speaker rushes the ending or the microphone clips on a loud syllable, replacing the vocal sound may carry those problems into the result. A short clean test gives you a practical basis for deciding whether to re-record, adjust settings or proceed with a longer passage.

What it does

  • Convert your own spoken performance into an authorised target voice.
  • Compare the character of two versions of the same line.
  • Preserve the pacing of an existing performance.
  • Prepare a cleaner source recording for the conversion.
  • Split a long take at pauses before processing individual sections.

Install and use

  1. Install the skill and official ElevenLabs CLI or SDK. Set an API key with credits and permission to use the selected target voice.
  2. Ask “Convert this line that I recorded myself using the chosen stock voice. Keep a separate audio file for review.”
  3. Compare the two performances and re-record any part whose timing or expression needs a change.

FAQ

Source

Made by ElevenLabs. Original skill: voice-changer, published on GitHub under the MIT licence.

View on GitHub