How to clone a voice and sing an AI track with it

Open Voice options in the studio, expand Clone a new voice, name it, upload 15 to 30 seconds of clean solo singing as MP3 or M4A under 10 MB, tick the consent box and click Clone voice · $5. The clone is created through Mureka's voice service, saved to your studio, and from then on it appears in the Cloned voice select. A cloned voice overrides the Female / Male buttons on every track you use it on.

Steps

1. Record or export a clean sample

The clone copies whatever is in the sample. Record one person singing alone, close to the microphone, in a quiet room, with no backing track, no reverb and no harmonies. Sing the way you want the track sung: a breathy sample gives a breathy clone. Keep it between 15 and 30 seconds and export as MP3 or M4A.

2. Open Voice options and find the clone section

Under Voice, click Voice options. At the bottom of the dialog is Cloned voice (overrides Female / Male) with a select, and under it a Clone a new voice section you expand by clicking it.

3. Name the voice and choose the file

Type a short name such as "Maya" or "My voice" in Voice name. Click the file field and choose your MP3 or M4A. Files over 10 MB or in other formats are refused with a message.

4. Confirm you have the right to clone it

Tick I own this voice or have the singer's permission to clone it. The button stays off until you do. Club Ready Music refuses clones of voices you do not have rights to.

5. Click Clone voice · $5 and wait

The sample is uploaded and sent to the voice service. Cloning takes from a few seconds to a few minutes; the status line next to the button tells you when it is done. The new voice is added to the select and chosen automatically.

6. Set the rest of the voice direction

With a cloned voice selected, the prompt says "lead vocal in the provided cloned voice" instead of a gender. Your Character, Delivery, How much vocal, Accent and Tone chips still apply, so a cloned voice can be asked to rap, chant or sing hook-only lines. Click Done.

7. Generate and listen to both takes

Press Generate. Both takes use the clone. Compare them for likeness and pick the stronger one. The voice stays in your select for future tracks; use Delete next to the select to remove it.

Tips

• Match the sample to the job. Clone a sung sample for sung tracks and a spoken sample for spoken-word tracks. One clone rarely does both well.
• Dry beats processed. A sample from a phone voice memo in a quiet room beats a reverb-heavy studio bounce.
• Keep the language consistent. A clone from an English sample can sing Spanish lyrics, but the accent of the sample carries over.
• Set None to go back. Choose None (use Female / Male) in the select whenever you want the preset voices again. The clone stays saved.
• One clone, many tracks. The $5 is a one-off per voice; rendering each track costs the usual 1 credit.

Common problems

Clone voice failed. The usual causes are a sample shorter than 15 seconds, music under the voice or a file that is not MP3 or M4A. Re-export a clean 20-second solo take and try again.

The clone does not sound like the singer. Likeness depends on the sample. Use a louder, closer, drier recording with a steady pitch, and avoid samples with long silences.

My track used the Female voice, not the clone. Open Voice options and check the Cloned voice select is set to your clone, not None, then render again.

I want to remove a clone. Select it and click Delete. Tracks already rendered with it keep their audio.