Voice licences & attribution

Every voice in the catalogue is cleared for commercial use. Friendly names are ours; they do not change the licence of the underlying model, and they are not the names of the people who recorded the source datasets.

Tiers

• Naadly Core Fast, unlimited, the widest catalogue.
• Naadly Studio Higher-fidelity voices, plus Hindi and multilingual.
• Naadly Clone Your own voice, cloned from a short sample.

Models in use

Attribution text

Where attribution is required, credit the source dataset in your video description, app credits or documentation:

Voices synthesised with models trained on Multilingual LibriSpeech, VCTK, LibriTTS-R and CSS10 (CC BY 4.0). Kokoro-82M is Apache-2.0. OpenVoice V2 is MIT.

Sound library

The music beds, ambiences and effects in the sound library come from two places, and each clip below says which. Most are public domain (CC0 1.0): no credit is required and no permission expires. The rest were made with Stable Audio Open Small under the Stability AI Community License, which is free for commercial use while this deployment's owner earns under $1M a year — if you resell above that, use the CC0 clips or take Stability's enterprise licence. Every creator and source page is listed, so a buyer reselling the output can see where each clip came from.

Cloned voices are yours: we record who consented and when, and we do not reuse your reference audio for anything else.