The INLINGO workflow: How do we do voiceover?

We’ve already talked about the work that project managers, translators, and editors do at INLINGO, and now it’s the voiceover team’s turn. Julia Molostova, our voiceover department head, wrote about what you need for voiceover, why we don’t edit files when they’re sent to the client for approval, and how voiceover can sometimes cause a localizer great pain.

Hi everyone, my name is Julia Molostova. I have been working in voiceover for almost 5 years and am head of the INLINGO voiceover department. I lead a project from beginning to end, I know all the pitfalls and lifehacks that go along with voiceover work, and now I’d like to tell you about INLINGO’s voiceover process. So I broke the process down into 5 steps. I’ll go over each one in detail.

Table of Contents

Step 1. Clarifying project information.

During this step, we request all the project information from the client. The minimum we need to create an estimate is:

Number of characters in the game that need voicing
Number of words per character
Are there children’s voices in the game
Is the timing of each phrase important (Lip Sync, Strict Time Constraint, etc.)
Is the project urgent

This information allows us to calculate the voiceover budget accurately and set the appropriate time frame for the project. Usually, the more we know about the project, the more smoothly the process goes, which reduces the risk of exceeding the budget or time frame.

Step 2. Studying the script and casting

After the client approves the estimated budget, we request additional information about the characters. Ideally, this should be a script that contains the following things:

A separate list of characters (name, gender, character description, accent if any, age, any references to another movie or game character that allow us to define the tone of the recording).
The lines that need to be recorded. The preferred format is 1 tab for 1 character, but other variations are acceptable for small projects. If the client leaves comments on the intonation for each line and indicates time limits and how they want the audio files named, they deserve an Oscar for their script 😊

As an additional option, if we notice errors in the text, we can offer clients proofreading of their source script before recording. We prefer to do this before the task goes to the actors, because if the actors themselves make edits it can lead to additional costs, and if the actors don’t notice errors, it can necessitate re-recording, which extends the time frame of the project.

After all issues are resolved, the script is approved and considered final. Then we begin casting. We generally offer 2-3 options per character, but there are nuances here, too. For example, a client might single out the main characters and request specific casting only for them, letting us choose for the secondary characters.

Step 3. Developing tasks for the actors, recording

Once the client approves all the castings, we start to create the tasks for the actors. It’s not enough to just give them directions on what parts of the overall script to use—everything needs to be separated by actors and by characters, and we have to put together separate mini-scripts. This part doesn’t take a lot of creativity, but it does reduce the chances of actors mixing up lines to zero.

*This is an example of a character description in a script.*

Step 4. Quality assurance, getting the client’s approval on the un-edited recordings

After the actors send the completed recordings, we double-check them. Namely: we listen to all the lines for background noise, check for any other sound defects (for example, an actor accepted a task, but didn’t warn us that she had a slight cold, resulting in what was supposed to be a girl’s voice sounding like an older girl with an “accent”). We also check the recording against the script to avoid any discrepancies. As soon as the recording passes this first wave of checks, we send the result to the client. We purposely don’t process or cut files during this step to avoid wasting time. So, if the customer wants to correct something about a character’s tone of voice, we just promptly get in touch with the actor and re-record this moment. This is much faster than if we process the file first and then send it to the customer for approval. And once again, this can all vary based on an individual project, and the exact sequence can always be discussed with the client.

Step 5. Processing, delivering

Once the client has approved all the recordings, we start processing, cutting, and renaming them. This is also where any plugins are applied (for example, a walkie-talkie or tunnel echo effect), voices can be put together in a chorus (for example, if several people are speaking at once – a crowd on the street, or, like in one of our past projects, a group of sages who all spoke at the same time).

*We use Steinberg Cubase 9 Pro to apply effects.*

“Wordless” Pain

As with any other process, voiceover has those moments that you would rather avoid, but everything would be just too boring and perfect without them.

When clients request an estimate without providing any information about their project.

It’s like a shot in the dark… Of course, we can give an approximate estimate based on the most common parameters and rates, but voiceover has so many nuances that expectations and reality can often be very different. Simply put, it is in our best interests that the final product is of high quality, so we need to be absolutely sure of the time frame and budget that we assign to a project.

There are time limits for a line, but the text doesn’t fit.

If we are the ones who prepared the text, we can figure it out. But when the script comes to us already finished, approved by the client, and then during voiceover we realize that we can’t make it fit the timing… Well, what doesn’t kill us makes us stronger 😊

Preferences when it comes to recording actors.

When the game needs a Colin Firth-esque voice, and you can definitely hear the right qualities in the recording that the actor sent in, but the client says that it’s not Colin-y enough, you’re suddenly struck by flashbacks of the meme about whether the dress is black and blue or white and gold.

*Your face when the client thinks you’re not Colin Firth-y enough.*

That’s why we try to ask clients for specific audio references for the characters so that the actors can get it right. Actually, it can be like a fun quest—sometimes everyone on the team gets together and listens to these lines, with everyone sharing their opinions about a controversial recording. And here we go! We’re looking forward to your feedback on our work, suggestions for new topics, and voiceover orders at —in whatever order you prefer. And may your project’s voiceover be perfect!

Workflow