The worked example, and the rate that came from somewhere else
The best passage on this page does the animation year out loud — twenty episodes at a session rate, totalled, and then brought back to prove that a good year on paper still misses the health threshold. The arithmetic is fine. The input is wrong.
The figure we used is a day rate from the network television code, the same number that appears elsewhere as a basic theatrical scale day. It is not a television animation per-episode minimum.
The figureWhere it actually comes from
$1,246 an episodeOur anchor number
No
It traces to a day performer rate under the network television code, and separately appears as a basic theatrical day scale. Not an animation minimum, and not per episode.
The nearest real figureNon-dramatic, half hour, one performer
Close
Around $1,223 from July 2025 and $1,260 from July 2026 under the same code. Similar in size, different in category, and not what we were describing.
And the agreement itselfThe one we cited
Expired
The animation agreement we were working from ran out on 30 June 2026. A new television and theatrical agreement was ratified in June and runs to 2030, so the animation minimums need re-sourcing entirely.
We are leaving the structure of that passage exactly as it was, because the structure is the good part. Doing a year out loud — a plausible number of episodes, a session rate, a total, and then the same total set against what the health plan actually requires — is the clearest thing on this page and it does not depend on any one figure being right. What it does depend on is the figure coming from the agreement that governs the work being described, and ours did not. We are re-sourcing it from a current animation rate sheet before that example goes back up. Until then the shape of the argument holds and the total does not, and we would rather say so than leave a confident number sitting there.
The number we derived, and the table we did not know existed
This page told you plainly that one figure was ours rather than the union’s. That honesty was right and the reason for it turns out to have been unnecessary.
We took a confirmed increase, applied it to an older four-hour day, got roughly $1,270, and said in the text that we had derived it rather than quoted it. The union publishes the actual table.
What the wage table says.
For the current period, off-camera day performer, up to three voices in a four-hour day: $1,135.
Our derivation ran about a hundred and thirty-five dollars high. The increase evidently did not apply as a flat multiplier to that line.
And a detail worth knowing because it looks like an error and is not: the on-camera day rate is identical to the off-camera three-voice four-hour rate. That is how the agreement is built.
Two things follow, and one of them is uncomfortable. The comfortable one is that flagging the derivation worked exactly as intended — we told you it was ours, it was wrong, and now it is fixed. Most pages would have printed a derived figure as a fact and never looked again. The uncomfortable one is that we should not have had to derive it. SAG-AFTRA publishes a wage table for this agreement on its own site, covering the whole term, and we did not find it. Our research thought a payroll company’s listing was the only public source and told us to verify it at the union. We did, and the union had it all along, to the penny. The lesson is not about voice rates. It is that “nobody publishes this” is a claim that needs the same checking as any other.
Forty per cent of what
We were handed a striking figure and told it deserved a paragraph of its own: that an employer pays a forty per cent benefit contribution on a digital-only campaign. It is real and it is not a contribution rate.
The actual provision is an allocation rule. Where a commercial’s only covered use is in the traditional digital category, a forty-sixty split is applied between covered and non-covered services — so only forty per cent of the compensation is treated as attracting contributions at all.
The direction is the opposite of the one we were about to publish. We would have told you an employer pays substantially more on a digital-only campaign. What the rule actually does is treat most of that compensation as outside the covered services, with the standard rate then applied to the smaller slice. That is a materially different fact and it points the other way. The standard commercials contribution is twenty-three and a half per cent, and a signatory advertiser or agency operating under the joint policy committee’s national agreement pays a reduced nineteen point nine five. Both of those we can stand up. The forty stays on the page as what it is — a split, not a rate — because it still tells a reader something real about how digital work is treated.
The contract year everybody got wrong
This page dated the commercials contract to a single year. Our research corrected it to a different single year. Both are wrong in the same way.
It is a three-year agreement running from April 2025 to March 2028, with wage steps built in at five per cent, then four, then three. What our research spotted was the second step, not a new contract.
And there is a wrinkle that explains how everybody keeps getting this wrong. The union’s own current-contract page labels the document with a one-year effective range, while the contract text inside it says three years commencing April 2025. So a reader checking the source carefully is given two different answers by the same organisation on the same page. We were wrong, our researcher was wrong in a different direction, and the label that misled both of us is the union’s own. The practical consequence for you is small and worth knowing: the figures step up on the first of April each year without a new negotiation, so a rate you looked up last spring is a year behind rather than out of date.
British rates, at last
Every figure on this page has been American. Our research called that the biggest structural gap in the best chapter we have and could not close it. We have closed it.
Equity publishes an audio rates document, last updated in July 2023. The union’s own website would not load for us at all — it has been unreachable from here on every attempt across several chapters — but the document is mirrored by the Casting Directors’ Guild and it is Equity’s own.
The workWhat the agreements say
Television and online commercialsThe IPA and Equity agreed document
£200
A basic studio fee of two hundred pounds an hour, tags at ten per cent of it, and then use fees on top calculated from audience measurement through Equity’s own online calculator.
BBC radioTheir audio agreement, 2022
£294
A full day at two hundred and ninety-four pounds for two transmissions, or two hundred and one for one. Part days at two thirds. Overtime by the quarter hour.
Commercial radioBanded by audience
£30–855
Scaled by listener reach, from thirty pounds under a quarter of a million listeners to eight hundred and fifty-five above five million. Digital priced per three hundred thousand impressions.
Video gamesOne company agreement, 2021
£175–300
By budget tier, hourly. The first hour is paid double, which is how the perpetual buyout is built into the rate. Equity has this agreement with one games company and no sector-wide deal.
AudiobooksPer finished hour
£130+
An Equity-recommended minimum around a hundred and thirty in studio and a hundred and fifty remote, with the open market running from a hundred to four hundred and up. There is no British royalty-share equivalent — the per-finished-hour model dominates.
One structural contrast worth having beside the American figures, because it is startling. Joining the American union costs a flat initiation fee of over three thousand dollars, plus flat base dues and a percentage of work. Equity’s annual subscription is banded by what you earn — reportedly starting under two hundred pounds a year below forty thousand of industry income and rising in steps for higher earners. One union charges everybody the same to join and the other charges you in proportion to what the profession has actually given you. We have that banding from secondary sources rather than from Equity, for the reason above, so treat the figures as indicative and go and look. The structural difference is the durable part and it is not in doubt.
Residuals or buyouts, and why the answer is both
This is the pivotal question for a British reader and the tempting answer is wrong in both directions.
It is not that America uses residuals and Britain uses buyouts, and it is not the reverse. Britain is split down the middle of its own market.
Unionised broadcast work
The modelSession fee plus use fees, calculated from audience figures.
WhereCommercials under the agreed document, and the broadcasters’ own agreements.
What it meansStructurally the same shape as the American residual system. The work keeps paying while it keeps running.
Everything else
The modelBuyouts. One fee, extended or unlimited use, no further payment.
WhereCorporate, e-learning, explainer, most games work, most of the open market.
What it meansThis is the larger half by volume, and it behaves exactly like the American non-union market.
The thing to actually watch
The mechanismA buyout is one payment for rights that may run for years across territories you never discussed.
The multiplierWorldwide usage commonly runs around three times a national rate, and a perpetual unlimited buyout carries the highest premium of all.
The trapThis is how a beginner accidentally sells a national campaign for a couple of hundred pounds. Ask what the usage is before you quote.
Which is why the games agreement above is worth reading closely even if you never work for that company. Doubling the first hour to cover a perpetual, all-territory buyout is the whole concept made visible in a rate card. Somebody sat down and priced what unlimited use is worth, and the answer was one more hour.
Can you live on it
The rate card and the bank statement are different documents. This is the second one.
The union figures reported in 2023 are still the clearest picture anyone has published, and they are three years old. Seven per cent of members cleared eighty thousand dollars. Fourteen per cent earned enough to qualify for health cover. The threshold has since risen and there is no published evidence that earnings did.
What the aggregators say
One large job site currently shows an average of just over a hundred thousand dollars, while the same site’s figure appears elsewhere as around eighty thousand, and it states a range from eleven thousand to a hundred and sixty. Three numbers, one source.
What the government says
The federal occupational data covers actors as a single category, with no voice-specific figure at all. There is no published median for this job.
What the surveys cannot see
Everybody who bought a microphone, paid for a demo, auditioned for a year and booked nothing. They are in nobody’s data, which is why every income figure in this profession reads better than the profession does.
And a whole column we never hadEvery net figure on this page ignores tax. For a self-employed British voice actor, equipment, acoustic treatment, software, platform commission, coaching, demo production and an apportioned share of home costs are ordinarily deductible business expenses, and there is a simplified flat-rate scheme for home working. We are describing the shape and not the numbers, and this is not tax advice.
The counterweight, compared honestly
Our argument that synthetic voice took the ladder rather than the top survives, and it is stronger with the buyer-side figure attached. The comparison usually made alongside it is not fair and we are going to say why.
Twenty-one per cent of voice actors reported knowingly losing work to a synthetic voice in the 2026 association survey, up about six points on the year before. And only twenty-six per cent of clients report having tried an AI voice at all, according to a platform’s own trends report — of whom seventy per cent used it on fewer than a quarter of their projects.
Two caveats, and the second one matters more than it looks. The buyer figure comes from a platform that sells human voice bookings, surveying its own paying customers, so a low adoption number among its own clients is not neutral market data and we are attributing it as theirs. And the cost comparison everybody reaches for is not comparing like with like. A generation subscription works out at roughly twelve dollars an hour of raw output. A human narrator’s per-finished-hour rate runs from a hundred and fifty to five hundred. But the human figure is fully loaded — it includes performance judgement, direction, retakes and the editing that turns a recording into a deliverable, and per the audiobook arithmetic below, a finished hour is roughly six hours of work. The twelve dollars buys none of that. It is an output price against a service rate, and quoting them side by side without saying so is the commonest sleight of hand in this argument.
None of which contradicts the original point. It quantifies why the ladder was the part that could go. The categories where nobody is directing anybody and nobody is paying for nuance are exactly the categories a beginner used to learn on.
The audiobook arithmetic
The best worked example we have that survives checking, and one detail about the recent royalty change that nobody is reporting.
A ten-hour finished audiobook is roughly a sixty-hour job. The industry ratio is about six hours of recording, editing and checking for every finished hour, which puts a ten-hour book squarely inside the fifty to seventy hours this page has always claimed.
And on the royalty change dated 26 May 2026, there is a detail worth having. The new pooled model pays a headline fifty per cent exclusive and thirty per cent non-exclusive, replacing forty and twenty-five, tied to a seven-year exclusivity commitment. The headline percentages went up. But narrators and authors are reporting real-world royalty drops of thirty to forty per cent once their titles are pooled, which is the opposite of what the numbers suggest. We are reporting that as what it is — accounts from people affected rather than an audited figure — because a seven-year commitment made on a headline percentage is exactly the kind of decision this page exists to slow down.
Grading the sources
The best-sourced page in the Academy needed four corrections, and two of them were to figures our own research handed us.
The claim, and where it came fromWhat we found
The animation episode rateOur anchor worked example
No
A day performer rate from the network television code, not an animation minimum. And the agreement we cited expired on 30 June 2026. The example is paused until it is re-sourced.
Digital-only campaigns carry a forty per cent contributionOur research, as a headline finding
Misread
It is a forty-sixty split between covered and non-covered services, so a smaller share of the fee attracts contributions. The direction is the opposite of the one we were given.
The commercials contract runs one yearOurs, and then our researcher’s
Both wrong
Three years, April 2025 to March 2028, with built-in annual steps. And the union’s own page carries a one-year label contradicting its own contract text.
The union does not publish this rateOur reason for deriving one
No
SAG-AFTRA publishes a wage table for the whole agreement term on its own site. We did not find it, and our derivation ran a hundred and thirty-five dollars high.
The published interactive day rateThe union’s own wage table
Primary
$1,135 for up to three voices in a four-hour day, current period. A payroll company’s listing matches it to the penny, which is a good sign about both.
British audio ratesEquity’s own document
Primary
Session fees, banded radio rates, a games agreement and an audiobook minimum. Their site is unreachable from here, so we read it through a casting body’s mirror of the same document.
Britain uses residuals rather than buyoutsThe tempting simplification
Both
Session plus use fees on unionised broadcast work; buyouts across the larger open market. Flattening it either way would mislead a reader negotiating.
Twenty-one per cent losing work to synthetic voiceThe 2026 association survey
Good
Up about six points year on year, corroborated in independent reporting. The clearest figure available on this.
Only a quarter of clients have tried AI voiceA platform’s trends report
Interested
Real, and it surveys that platform’s own paying customers, and the platform sells human bookings. Attributed as theirs rather than as market data.
Twelve dollars an hour against a hundred and fiftyThe standard comparison
Unfair
An output price set against a fully loaded service rate. The cheap side buys no direction, no retakes and no editing. We are printing both with the caveat attached.
The pooled royalty pays moreThe headline percentages
Contested
Fifty and thirty replacing forty and twenty-five, and affected narrators reporting real drops of thirty to forty per cent. Reported as accounts, not as an audited figure.
Confident rate figures on a shopping domainRanking high on this query
Fabricated
Attributed to an unnamed analyst on a social network. The union publishes no voice-only membership count and no such report exists. Nothing on that page is traceable.
One thing we could not do at all, for the fourth chapter running. Equity’s own website will not load from here. Every British figure above comes from their document as mirrored elsewhere, or from secondary reporting, and we have said which is which. If a number matters to a decision you are making, go to the union directly.
Where we differ from the standard account
What does a voice actor earn?
Commonly saidAn average salary, quoted confidently.
What we foundOne large job site publishes three different figures. There is no published median for this job at all.
Does the union publish its rates?
Commonly saidHard to find, so people derive them.
What we foundA full wage table on their own site, covering the whole agreement term. We derived a figure we did not need to.
Do digital-only campaigns pay more into benefits?
Commonly saidForty per cent, quoted as a rate.
What we foundIt is a split, not a rate. Less of the fee counts, not more.
Does Britain use residuals or buyouts?
Commonly saidOne or the other, depending who you ask.
What we foundBoth. Session plus use fees on unionised broadcast, buyouts across the larger open market.
What does a British session pay?
Commonly saidNobody publishes it.
What we foundEquity does. Two hundred an hour as a basic studio fee for commercials, plus use fees on top.
Is AI ten times cheaper?
Commonly saidTwelve dollars against a hundred and fifty.
What we foundAn output price against a service rate. The cheap side buys no direction, no retakes and no edit.
Did the audiobook royalty go up?
Commonly saidFifty per cent, up from forty.
What we foundThe headline rose and narrators report real income falling once pooled. Seven years is a long commitment.
Do one thing tonight. If you are quoting for a job this week, ask what the usage is before you name a number. Territory, media, duration. It is the single question that separates a session fee from a buyout, and it costs nothing to ask. Everything else on this page is a rate card. That question is what decides which one applies to you.
