Timbre voice foundry

A foundry for the human voice

A speaker sits one session in our booth. We cut the model from that recording and nothing else, and the speaker can pull it back.

Cut in-house, Rotterdam Six voices in the open catalogue

The catalogue

six entries, open licence

Each card is set the way the voice reads. Weight, width and colour are the specimen; there is no waveform here because a waveform tells you nothing about a voice you are about to cast.

How a voice gets cut

booth to catalogue
  1. The speaker sits a session

    One booth, one supervising engineer, and a script written against the range you need rather than a generic passage. The speaker is in the room and reads the terms before the first take, not after.

  2. We cut the model in-house

    Training runs on our own hardware against that session and no other audio. When it is done, the speaker auditions their own voice and can send it back to be cut again.

  3. You direct it, they hold it

    You get the voice, the direction controls, and a written licence. The speaker keeps a revocation they can exercise themselves, and every render carries provenance metadata on the way out.

Consent is the product

Voice cloning goes wrong in three places, so the terms are written against those three and are the same terms for every entry in the catalogue.

Provenance
Every voice names the session it came from. Nothing enters the catalogue from scraped audiobooks, podcasts or films, and a voice with no traceable session does not get cut.
Revocation
The speaker can withdraw their voice from the catalogue at any time from their own account. Withdrawal stops new renders immediately; existing licences enter a wind-down we publish in full inside the terms.
Declaration
Renders leave with signed provenance metadata attached. Strip it and the licence lapses. We would rather lose the work than ship a voice that cannot be identified as cut.

Full terms and the speaker agreement are published before you are asked for anything.

Book a session

Tell us what the voice has to do and we will come back with a booth date, a read of whether an existing catalogue entry already covers it, and a rate. Nothing is charged before the terms are signed.

We reply from a person, once.

Changes which engineer picks it up.

You are not on a list. This goes to the studio inbox.

Logged. An engineer picks it up from the studio inbox.