How to create a voice clone

AI Voice Cloner is for voices you have permission to clone. By the end of this guide, you will have an authorized clone saved in My Voices, ready to test in Text to Speech before using it in a final project.

Before you start

• Confirm whether Personal or a Workspace is selected before uploading source audio. The completed clone belongs to that selected space; a Workspace clone is shared with its members.
• Confirm you own the voice or have explicit permission to clone it.
• Prepare a short, clean recording with one speaker and minimal room noise.
• Choose the language spoken in the recording before submitting the clone.

Create it in Giggy

1. In the left sidebar, open AI Voice Cloner.
2. Upload clean speech audio from a voice you are allowed to clone.
3. Enter the voice name.
4. Choose the language.
5. Select Generate.
6. Review the generated clone in My Voices before using it in Text to Speech.

Improve the result

• Use clean speech with one speaker and minimal background noise.
• Do not use music, overlapping speakers, echo-heavy rooms, or heavily processed audio.
• Use 3 to 10 seconds of clear reference speech when possible.
• Record in a quiet room with the microphone close enough to capture speech clearly but not so close that plosives distort the audio.
• Use a natural, expressive read instead of a flat sample so the clone has useful cadence and tone.
• Do not clone or imitate a real person without authorization.
• Preview the clone with a short script before using it in a final project.
• After the clone is created, add expression tags in Text to Speech only where the script needs the cue.

Examples

• Good source audio: one speaker reading naturally in a quiet room for 3 to 10 seconds.
• Poor source audio: music under the voice, overlapping speakers, heavy echo, phone noise, or a clipped recording.
• Good reference audio: use 3 to 10 seconds of clear speech from one speaker without music or echo.

Common mistakes

• Uploading audio you are not authorized to clone.
• Using a noisy, compressed, musical, or multi-speaker clip as the source.
• Skipping a short test script before using the clone in production.

Use cases

Authorized personal voice

• Create reusable speech from your own authorized voice.
• Generate draft reads when the approved speaker is not available to record.

Client and team voices

• Build client, performer, or employee voices only when permission is documented.
• Keep a consistent approved voice across multiple scripts, campaigns, or repeat content formats.

Voice consistency

• Use one approved voice consistently across a series of scripts or campaigns.
• Use short test scripts before using a clone in a final production workflow.

Open AI Voice Cloner · Creative app