Voice, AI & Synthetic Use
Last updated 14 August 2026
The short version. Booking me buys the usage we agree in writing, and nothing else. My voice may not be used to train an AI model, to create a synthetic or cloned version of me, or to generate new speech in my voice — unless we have agreed that separately, in writing, with separate payment.
1. What a standard booking includes
When you book me for a project, you are licensing a recording for a specific, agreed use: a particular script, in a particular place, for a particular length of time. The quote is based on that use. Anything outside it is a new conversation, not an assumption.
2. What a standard booking does not include
- AI training. Recordings may not be used to train, fine-tune, or form part of a dataset for any machine-learning or speech model.
- Voice cloning. No synthetic replica, digital double, or voice model may be created from my recordings.
- Synthetic generation. No new speech may be generated in my voice, whether or not it resembles the original recording.
- Resale or sub-licensing of the recording to a third party.
- Use beyond the agreed territory, platform or period.
3. Auditions
Auditions and custom test reads are provided free, for one purpose: so you can decide whether to book me. They are not licensed for broadcast, publication, internal use, or AI training of any kind. If you would like to use an audition recording, please ask first.
4. If you do want synthetic use
Ask me directly. Some projects have a legitimate reason to want a synthetic option, and I would rather have an honest conversation than discover it afterwards. Any such use requires a separate written agreement covering scope, duration, and payment. The answer may be no. It will always be a straight answer.
5. Automated crawling of this site
The words on this website may be read, indexed and quoted by search engines and AI assistants — that is how people find me. The audio files are different. This site's robots.txt asks AI crawlers not to access the audio directory, and the audio is served with headers signalling that it is not available for AI use.
To be straightforward about it: those are requests, not walls. They stop well-behaved crawlers and they create a clear, dated record that permission was refused. They cannot physically prevent someone determined to ignore them. Anyone who tells you otherwise is overselling.
6. Why this page exists
Not because I think AI is going away. It is getting better, and pretending otherwise would be silly. But a voice is not a stock asset — it belongs to a person, and how it gets reused should be that person's decision. Putting it in writing is simply the professional version of that.
7. Questions
If anything here is unclear, or your project needs terms different from these, please just ask before booking. It is far easier to agree scope at the start than to unpick it later.