Synthetic voice is now a line item on enough Australian TVC briefs that every producer needs to know how the consent layer works. The MEAA Voiceover Standard Contract + AI Rider 2024 is the mechanism. This guide covers what the rider does, when it applies, how it interacts with the base voiceover fee, and the liability exposure if a producer ships synthetic voice without one.

What the AI Rider actually is

The MEAA Voiceover Standard Contract + AI Rider 2024 is an add-on to the standard MEAA voiceover contract. The base contract licenses the human performance — the voice take the performer delivered in the booth. The AI Rider is a separate consent instrument that licenses AI-adjacent uses of that performance: training data, synthetic generation, cloning, voice-alike output and AI-assisted manipulation beyond standard post-production.

It is structurally independent from the rate-card hard-dollar cells published in the MEAA Voiceover Rate Card 1 July 2025. Paying the Visual, Audio or VO Subs cell for a human read does not license an AI use. That is a separate deal, on separate paper.

The base voiceover fee does not license AI use. The MEAA Voiceover Standard Contract + AI Rider 2024 is the consent framework, and it is negotiated separately.

When the rider applies

The rider engages when the brief moves beyond a straight human voice take. In practice, that usually means one of the following:

  • Training data licensing— the producer, agency or a third-party vendor wants to use the performer's recorded voice to train a model.
  • Synthetic voice generation— fresh audio is generated by a model in the performer's voice, without the performer returning to the booth.
  • Voice cloning — a digital clone is created from the existing performance and then re-used in future campaigns, language versioning, dynamic creative or other variations.
  • Voice-alike generation— synthetic output designed to evoke a specific performer's voice, whether or not it was trained directly on them.
  • AI-assisted manipulation beyond standard post — model-driven timing, pitch, emotion or language edits that go past conventional post-production corrections.

Standard post — EQ, de-noise, level, conventional comp — does not by itself trigger the rider. The trigger is model-generated output or model-trained reuse.

The rider is written so consent is specific. A performer consenting to a clone for one campaign has not consented to the clone being re-used in a later campaign, repurposed across languages, or licensed to a third party. Each additional use is a fresh consent conversation and typically a fresh fee.

This matters most on library builds and AI-native creative pipelines. If the brief is "we want to clone the voice and use it for the year across six spots," the rider has to cover that specifically. "We'll figure it out later" is not consent.

The producer workflow

The cleanest way to run an AI-adjacent voiceover booking is to flag it upstream and budget for it as a separate cost. A practical sequence:

  1. Flag AI usage in the casting brief. The performer and their agent need to know before submission whether the brief involves cloning, training, synthetic generation or voice-alike.
  2. Price the human read first on the voiceover calculator using the MEAA Voiceover Rate Card 1 July 2025 base cell, territory multiplier and platform bundle.
  3. Budget the AI Rider separately. The rider fee is a negotiated amount on top of the base read and is scoped to the specific AI use requested.
  4. Get written consent before the booth session. If the performer is uncomfortable with the AI scope, that needs to be resolved before the session, not after.
  5. Document the scope precisely — what is being trained, cloned or generated, for which campaigns, over what term, across which territories.

Shipping synthetic voice without a rider

The liability exposure for skipping the rider is real. If a producer clones a performer's voice from an existing take and ships a spot without separate AI consent, the performer has an enforcement pathway through MEAA, and there is a likeness and personality-rights layer on top — Australian performers can pursue passing-off and misleading- conduct claims where a synthetic voice is presented as their performance without authority. Vendors that trained on un-licensed performer audio expose the producer as well as themselves.

This guide is a reference, not legal advice. AI-usage deals are still a moving area and the safest play is always to get independent legal advice on the consent language before a clone or a trained model goes to air.

Where Callsheet fits

The voiceover calculatorprices the human read cleanly against the MEAA Voiceover Rate Card 1 July 2025 — base cell, territory ladder, platform bundle, super on top, agency commission. The AI Rider fee lives alongside that number as a separate negotiated line. Keep them on the quote as two distinct items so the client, the agent and the performer can all see what has been licensed and what hasn't.

For the rate card itself and how the hard-dollar cells work, see the MEAA Voiceover Rate Card 1 July 2025 explained.