声匠 VoiceSmiths
Back to blog

Monkey King-Style AI Voices: Designed vs. Cloned

Want a voice that sounds like a famous character or presenter? The legal, commercial path is voice design — generate an original character in that style. Here is why cloning real or copyrighted voices is off the table, and how to write descriptions that land the style.

Aug 26, 2026VoiceSmiths TeamVoiceSmiths Team

Listen to this article

AI narration · about 3 min
0:00 / –:––

"Can you make the Monkey King's voice? The wolf from that cartoon? This famous presenter?" The honest answer has two halves: you can create that style — you cannot clone that person.

1. Why we don't clone real or copyrighted voices

  • Most jurisdictions treat a person's voice as a personality right; China's Civil Code (Art. 1023) protects it like a likeness, and courts have already awarded damages for unauthorised AI voice use.
  • A cartoon character's voice is an actor's performance — using it commercially needs a licence.
  • Provider terms forbid it, and accounts that try get banned — taking your whole product down with them.

Cloning is for your own voice or voices you hold written consent for. See voice cloning and consent.

2. Use voice design to create an original character in that style

Voice design needs no recording: describe the voice in one sentence and the model generates a brand-new voice that belongs to no real person. That is where our character library came from:

  • Monkey King — "high-pitched, cheeky, slightly raspy young male with opera flair, fast, boastful, lots of laughter"
  • Broadcast Announcer — "deep, resonant, perfectly articulated Mandarin with anchor-desk gravitas"
  • Cartoon Wolf — "scheming, raspy, slightly nasal middle-aged male, theatrical and comic"
  • British Butler — "polished RP, calm, dry, impeccably courteous"

Play them all on the homepage voice gallery; every character has its own portrait.

3. Writing descriptions that land

  1. Who: age, gender, role (storyteller, livestream host, late-night DJ).
  2. How: pitch, weight, pace, breathiness, accent.
  3. Where: opera stage, keynote, midnight radio — the model adjusts the mood.
  4. Generate three previews, keep the closest, tweak one adjective at a time.

4. Once you have it

A saved character works in text to speech, multi-speaker dialogue, long-form audiobooks — and in the voice changer, which re-skins your own read in that character while keeping the performance. The "Audio tools" tab on the homepage shows a livestream pitch turned into the Monkey King and the British butler.

In one line: style can be designed; people cannot be cloned.