Voice over mistakes: what derails a project, and the warning signs to watch for

The file arrives on time. The editor drops the French track under the picture, and something is off. The energy runs bright for the sequence, the read spills past a locked edit, and someone asks whether the rights cover the version going out in Belgium. Nobody answers.

The reflex is to blame the casting: wrong voice, book someone else next time. Occasionally that is true. Far more often the project was scoped too fast, weeks earlier, when everything still looked simple. Hiring a voice over looks like a simple purchase: pick a voice, send a script, get a file.

The voice over mistakes that cost real money are almost never made in the booth. They are made during preparation, and they surface only once the project is committed. That goes double when you run a voice over project in a language you do not speak, the everyday situation for a producer ordering a French voice from New York.

What is decided before the microphone is switched on

1. A brief that describes the deliverable, not the intention

“Female voice, warm, 30-second spot.” The brief fits on one line and says nothing about what matters. The actor receives a format, never a direction. Every take becomes a hypothesis: they try an intention, you listen, you correct, they go again. The back and forth is not a talent problem. It is an information problem.

Five elements change that, and they fit in one paragraph:

• where it runs: medium, channel, how long the audience is exposed to it;
• what the message has to do: inform, sell, reassure, train;
• who is listening, described by what they already know about the subject;
• three adjectives of intention, along the lines of “calm, instructional, reassuring”;
• what you specifically do not want to hear.

The last one is the most useful and almost always the one left out. “Not salesy” or “not too bright” rules out, on day one, the directions that would have cost a full retake.

A voice over brief that crosses a language needs one line more: which words stay in English, how the brand name is pronounced, whether the translated script may be reworked to fit the picture. Left unsaid, each becomes a decision the actor makes alone in the booth.

2. Casting decided on timbre

Voices get played in a meeting, the room keeps the one it finds most beautiful, it gets signed off. Taste is not the problem. The criterion is: a timbre wins a meeting room, an intention reaches an audience.

The booth disproves the myth of the ideal voice quickly. There is no universally good timbre, only timbres that serve a given project well: a low, settled voice carries a documentary, an unusual one makes a character exist. That gap cannot be judged from a generic demo.

The risk doubles when the recording language is not yours. A French demo, heard by an ear that does not speak French, can be judged only on timbre, because intention is carried by meaning. Have the voices you kept read the opening lines of your real script, and have those reads checked by someone who speaks the language. A voice catalogue is where a shortlist starts, not where a decision ends.

3. A script written for the eye and never timed

The text was drafted in a document, reviewed, approved. Nobody read it out loud. On session day it overruns the format, and something has to be cut. Under pressure you rarely cut what is superfluous: you cut what comes out most easily, often the argument that justified the film.

Text is spoken more slowly than it is read. Breaths take up room, a legal mention eats several seconds, an address slows everything down. Read the script out loud at the real pace of delivery and time it before you lock it, then keep margin. ScriptTimer gives you a first duration, and reading it out loud confirms it.

A script that crosses a language carries a second trap: a version approved in English rarely lands on the same duration in French. The translation comes back, everyone checks it for accuracy, nobody times it, and the overrun turns up with the studio clock running.

An experienced actor will suggest adjustments: shorter sentences, phrasing easier to articulate, breaths moved. Those edits are approved before the session, never on discovery of the delivered file. On a commercial voice over, where a few syllables decide the edit, that habit prevents most bad surprises.

4. Rights left for later

The quote talks about the voice. It says nothing about what you are allowed to do with it. As long as the campaign stays inside its original perimeter, nobody notices.

“We have never had a problem with this.” That is accurate, and it is precisely the mechanism: the subject does not come up in production, it comes up with success. The campaign works, it gets extended, it moves to another country, and you renegotiate after distribution from a weak position. Or your legal team blocks the launch, because nobody can establish what was authorized.

Four coordinates are enough to scope it, and they belong before the order:

• the media (web, TV, radio, internal, point of sale);
• the territories;
• the term of use;
• exclusivity, if your brand needs it.

Those four lines belong in the quote, not in an email thread somebody has to dig out two years later. One word travels badly here: buyout names a payment structure, not a scope, so write the four coordinates anyway. What a buyout covers, and why two quotes for the same voice are not comparable, is a subject of its own, treated in our article on buyouts and usage rights.

5. Group sign-off with no single decision maker

Several people listen, several opinions come back, none has the last word. A retake is requested to satisfy one comment, then another to undo it because a third opinion preferred the first. The actor receives instructions that cancel each other out. The final version becomes a soft compromise: nobody is against it, nobody is convinced by it.

International projects add layers: the brand, the local market, the agency, sometimes a distributor. Each has a legitimate view, and none closes the loop on its own. Name a decision maker before the session, in writing, on the same page as the brief. Opinions go to that person, who arbitrates and passes on a single instruction. Voice direction is decided, it is not voted on.

6. A schedule with no pickup window

The recording is booked the day before the launch. At that point there is no room to redo a line, no room to align the mix to the channel, none if the decision maker changes their mind on hearing the result. You go out with the only version you have.

Count backwards from the air date, laying out the real stages: script sign-off, casting, session, edit, mix per channel, final approval. Then add a pickup window, the only thing that lets you correct without arbitrating in a hurry. Add the time zones too: a comment sent at the end of a New York afternoon reaches a European studio the next morning, so a single round trip of feedback crosses a night.

The warning signs on the other side of the table

The second tipping point sits with your provider, and it reads before the order. A showreel shows craft. How a project is run shows up in the questions you are asked, and in what the provider puts in writing.

No question about where it will run

You send a script, you get a lead time. Nothing about the channel, nothing about the audience, nothing about what the message has to trigger. What was not asked will not be taken into account. A radio spot and an e-learning module have little in common, in pace or in energy, and a provider who never asks where the voice will land delivers a clean read calibrated on nothing.

No voice direction during the session

The actor records alone, at home, and sends a file. On utility content that can be enough. On a message that has to carry a precise emotion, the missing direction is paid for in retakes, then in compromises. Distance is not an excuse: sessions have been directed remotely for years, over video or a studio link, with the client listening in and a director adjusting in real time. Ask who will direct yours, how you can attend, and who is listening for the meaning rather than the sound.

The rights perimeter missing from the quote

One line reading “voice over”, a recording duration, a lead time. Nothing on media, nothing on term, nothing on territories. You believe you are choosing between two comparable services. You are choosing between two perimeters with nothing in common. And the day the use goes past what was authorized, the advertiser carries the risk, not the studio. That is what you expect from a French voice over agency used to working with advertisers: a perimeter written, dated and enforceable.

No pickups provided for in the contract

The quote does not say what happens if a line has to be redone. The question looks secondary at order time, and becomes a negotiation at the worst moment, when the schedule is tight. The distinction matters: a script change requested after the fact is a new order, while a take redone for a mispronunciation or a recording defect is on the provider. Have both written down, with the lead time attached.

The same delivery whatever the channel

One file, one render, for television, radio, web and internal training. Level and processing requirements differ from one channel to the next: a mix that sits right in one place sits wrong in another, and the adjustment lands on your editor the day before launch. This is where going through a studio becomes concrete: the render is calibrated for the target channel, and when several voices are recorded in different places, somebody answers for the consistency of the whole, so a series of modules sounds like the same room. The only question that concerns you is who that somebody is.

What is left to lock down before you order

None of these mistakes shows up on a quote. They show up later, when points that should have been settled in advance get decided in a hurry. All of them are handled upstream, with a few questions asked at the right moment and someone who answers them in writing.

Most of these mistakes are avoided in twenty minutes of scoping. Describe the project: the channel, the deadline, what the message has to trigger, what you do not want to hear, and the language it will be recorded in. Let’s talk about scoping before we talk about voices, and your quote request is the place to start.