Cast a synthetic voice, tune it, save it. Synthesized from scratch — the generator is given no voice sample, no donor and no reference audio.
Sign in to build voices. A new account gets ten builds to compare, free, no card. Creating one accepts the terms of sale and use.
Your voice is a file. To make it say something new, load it into any voice-cloning tool as the reference. Vopus does not run those tools and is not affiliated with any of them — this list is here because the question comes up, and because knowing which ones accept a synthetic voice saves you time.
Your Voice Authorship Papers are what clears the upload. Every one of these voice cloning models asks you to confirm you hold the rights to the voice. You do: it was synthesised from numbers, with no human recording, no donor corpus and no reference audio at any stage. Keep the Voice Authorship Papers with the file.
Every Vopus voice carries the Vopus mark. A signal below the threshold of hearing is written into the audio itself, carrying the Vopus identifier and your voice's serial number. It survives normalising, resampling, trimming and mp3 encoding, so the file can still be shown to be yours after it has been edited, converted or re-recorded elsewhere. It is provenance, not a restriction: it carries no name and no account detail, it does not call home, and it does not limit where the file can be used.
ElevenLabs Professional Voice Cloning and Speechify both verify consent by having the speaker read a phrase aloud on the spot. Nobody can speak in a synthesised voice, so there is no way to pass that step. ElevenLabs Professional also wants 30 minutes to 3 hours of audio.
XTTS v2 / Coqui carries a licence with revenue and user limits, and Coqui closed in January 2024, so there is no longer anyone to buy a commercial licence from. F5-TTS ships MIT code with non-commercial model weights, which is easy to miss. Fine for personal projects, not for paid work.
Terms change. Check the current ones yourself before you rely on any of this.