How do they localize voice acting in games?

How does the sound localization of Western games into Russian take place App2Top.ru Oleg Mironov from Logrus IT company told.

The report was first read in the walls of Scream School.

Features of the work

Voicing a game is a complex technical process, but all its stages can be completed painlessly if you have a good idea of all its components and prepare in advance for the pitfalls along the way.

Most often, it becomes known about the project with the need for sound recording in advance.

The term can be from 2 weeks to 6 months or more. This means that there is time to prepare for localization, evaluate your resources, the realism of deadlines and the prospect of passing the work on time.

It is necessary to evaluate all aspects – the availability of translators and editors, the loading of the studio with other projects, the availability of actors who will cope with the tasks.

If the project is large and the deadline is not far off, it is worth considering the need for parallel recording at several studios.

Workflow organization and team

In our team, each specialist performs his specific functions.

The project manager, most often one of the editors, organizes the functioning of working groups: issues texts to translators, receives and verifies translated texts from other editors; informs the sound group about the need to record certain materials, participates in the process of selecting actors for casting, controls the deadlines for translation.

The director of dubbing directs the recording process in the studio, supervises the work of the actors, the sound engineer.

The sound engineer controls the quantity and quality of the received English audio files, and then the quality of the localized audio files.

The sound engineer works in the studio during the recording, helps the dubbing director during the recording process.
The engineering group programs tools that help prepare audio for recording and keep records of recorded material.

The testers check the quality of the recorded audio.

Preparation in localization

Work on sound localization begins long before the very process of voicing. Good preparation is the key to successful and trouble–free localization.

The process begins with the fact that the publisher’s project manager sends English texts for dubbing for translation. We refer to the texts of the voice–over texts of the script part (English “scripts” – texts, speech of the heroes of the game), texts of minor characters.

The translators start translating, then the editors stack the texts under the original audio files.

A well-placed text for an audio file allows you not to be distracted during recording by editing the text and laying it under the original, thereby giving the maximum amount of time to sound recording. Actors correctly selected for the characters will be able to record the casting faster and more efficiently and then voice the game.

The sound engineer analyzes the received English files, compares their number with the script, prepares for sound recording, for example, assigns temporary attributes: serial number, date, recording type, so that the program can automatically correlate files with a certain character, actor or recording type.

Careful preparation allows you to record sound as efficiently as possible with a minimum number of duplicates and defects.

Translation difficulties

Localization of characters with regional peculiarities of the language

In all parts of one famous turn-based strategy, dwarves speak the Scottish dialect of English. The recording requirements even indicate that actors with a Scottish accent are needed.

In English, dialects are actively used to characterize characters. The option to transmit dialects of a foreign language in the existing ways in the Russian language does not work. It almost always turns out to be caricatured: attempts to convey the dialectal features of the character’s speech most often slip either to the conditionally Baltic, or to the completely anecdotal Caucasian. And the Scottish dialect in Russian is basically impossible to convey, even the British themselves hardly recognize it. Russian Russian localization, the actors play dwarfs, but in Russian.

In one major AAA project, Edinburgh-born Alexander Bell, the one who invented the telephone, was voiced in the English version by an actor of Scottish origin Mark Rowley. In the Russian localization, this was done by actor Dmitry Kurta without a hint of any dialect, since in this case the transfer of the dialect in the localization would not carry any semantic load for the player.

A good example of accent transmission can be found in the same game. The German philosopher and author of “Capital” Karl Marx was voiced by Alexander Novikov. A slight German accent, not on every word, but sometimes cutting through, individual German words, and here it is a complete image. It could not be done, but the German accent is more familiar to a Russian person, in the game we meet Karl Marx in England and with the help of an accent, the fact that he is a foreigner is additionally emphasized.

The publisher may ask Russian-speaking actors to voice English texts in English with a Russian accent. Most often this is done to obtain samples, which English-speaking actors will subsequently use when reproducing a Russian accent.

F*** you, f*** you! F***! Who the f*** are you? – localization of obscene language

Some time ago, there was a lively discussion in the media about the permissibility of a mat in the cinema to create realistic artistic images. I personally doubt that obscene language somehow helps in this, and an example of this is the wonderful films of Soviet times in which actors created memorable images of colorful characters without a single obscene word.

In our localizations, we do not allow the use of obscene expressions. The Russian language is extremely rich in expressive means to resort to such extreme ways of conveying strong emotions and psycho-emotional states. In addition, replacing “fuck” with a Russian obscenity is a stylistic discrepancy, because this swear word is much softer and more freely used.

Working with the family

– F*** you!
– Fuck you!

This example of translation is an illustration for another of the difficulties that can often be solved, alas, only with the help of developers. Russian Russian is associated with differences in the grammatical structure of the English and Russian languages, namely, the presence in the Russian language of the genders of nouns and the absence of such in English. It is most often found when localizing texts of non-player characters.

For example, a character in English has a string “Screw you!”. If we don’t know the gender of the addressee, then the first thing that comes to mind is to translate the line as “Fuck you”. It is better to solve the gender problem already at this stage and choose a neutral option, for example, “Fuck you”.

A more difficult case is when the hero can be of any gender. In this case, substantive work should be carried out on each specific case: to look at the text, what can be done “asexual”, what cannot be sent to developers so that they are divided into 2 lines.

Casting of voice actors

While the translators are translating the texts and looking for a solution to the difficulties described above and the like, English sound files are being sent to the studio.

Work begins with sound. Most often, localization of a large game intended for simultaneous release in several languages goes in parallel with its development. And this brings its own nuances to the sound localization process. For example, the developers changed a significant part of the translated and voiced texts and sent an updated script with the words “here, figure out what’s new and what needs to be redone.”

The recording itself begins with the casting. When the publisher finishes dubbing in English for the main characters at his studio, he sends English samples with the requirement to cast the actors. There are two types of casting: “live” (live) and casting based on votes.

For a “live” casting, we invite actors to the studio to perform and record a certain passage from the text of the game. As a reference, they usually use either special samples, or videos, or footage from mocap filming from the pavilion. The customer can be present at the live casting, most often remotely via Skype.

For casting, samples of the voices of those actors who, in the opinion of the project manager and the sound team, will be able to best play the characters of the game are selected from the voice database. The database consists of audio files of actors selected from previous projects. As a rule, three actors are selected for one character, sometimes more – depending on the requirements of the customer. This applies, for example, to those main characters for whom there was no live casting.

An important element of the work is the selection of the right samples to send to the publisher for the selection of actors. The basic principle is that the sample should correspond to the English original by genre, that is, if a warrior shouts in the original, then there should also be a scream in the sample, if there is an elder, then the sample should correspond to the original.

In the process of work, we often faced the problem of a shortage of suitable samples. As a result, we have developed a system of so-called “ideal” casting. We have prepared materials for a 30-minute sound recording, which includes the most common types of heroes of different ages and all the actors who come to us record samples of different types and ages, so we can quickly find the right sample. At the same time, we evaluate how flexible the actor is, i.e. he can play different characters, change his voice for different ages.

Working with actors

We are constantly looking for feedback on our work and in general reviews of projects with voice acting and not only on games, and very often we encounter complaints from viewers and players that “the same voices are all around.” A common problem in the media industry, in our opinion, is the lack of professional dubbing actors who, in addition to talent, also “know how” to work efficiently and quickly. As I have already mentioned, usually sound localization takes place in a short time and in parallel with the creation of the game itself. Therefore, the voice directors invite those who can do the project well, who you can rely on.

We are constantly looking for new voices and for each project there is a rule to attract up to 15% of new actors, that is, those with whom we have not worked yet. That is, every 6th actor in a new project is a new one. Why so little? The fact is that each new person is a certain risk and it is this amount that allows you to cover the risk and, in extreme cases, rewrite the “unsuccessful roles”.

Usually we start working with a new actor with episodic roles, with the so-called “grenades”. Grenades can also be “shouted” without enthusiasm, and such a hack is immediately visible. And if we see that the actor is trying, there is potential, then we take note and invite him again.

The actor must be in the profession, be in Moscow and be able to come to our studio to record, so we eliminate the risk that the recorded files will differ in quality from those recorded by us. So if an actor/actress combines sound recording with filming, then he/she can easily go to another city and we will have to rewrite the role, etc.

Not every talented actor can voice games, this applies to movies and cartoons. I will say more, not every voice actor can voice games. When voicing games, the actor is faced with the task of getting involved in the role for a short time, understanding it and playing it, often without having a visual image in front of him, but only hearing an English sample. Usually we get character portraits only for the main characters.

The concept of effective recording

In order for the recording to be as effective as possible, you need to have texts ready for it.

By ready-made, I mean those that were not only correctly translated into Russian, but also laid down to the sound. The latter is handled by a stacker editor who brings the length of phrases in Russian to the original meaning. This is done in Excel, which allows you to control the length of phrases in Russian and English.

Example of laying:

  • English text: “Freeing future generations from a city ruled by Templars. You know, Jacob Frye, you might just be right.»
  • Translator’s text: “Liberate future generations from the city under the rule of the Templars… You know, Jacob Fry, maybe you’re right.”
  • The edited text, laid down under the length: “To free the descendants from the city under the rule of the Templars. You know, Jacob Fry, maybe you’re right.”

In our understanding, the recording should be conducted with a minimum number of duplicates. The dubbing director should be able to control the actor, guide him/her on the right path.

The process of voicing

After the casting is approved, the actors are called and confirmed that those who voice the main characters are ready to be on the record during a certain period.

The recording is conducted in a program convenient for the customer, for example, if you need to submit material only in the ProTools program and that’s it, then there are no options, you need to adjust.

Our task is to return the recorded materials in a form ready for integration into the game and meeting the stated technical requirements and on time.

The translation is also carried out in the program in which the customer sent the materials.

There are various translation programs that help the translator and speed up translation, for example, MemoQ, Trados and others. But, despite the existence of advanced tools, very often texts are sent to Excel. This program also allows you to keep records and prepare files for recording. It provides very extensive opportunities for organizing materials, accounting for received, recorded files, intermediate totals, and built-in Visual Basic tools allow you to create utilities specifically for your needs.

The main thing is that the material you have to work with has something to cling to, that is, there would be elements that allow it to be systematized. For example, so that the file names have a stable structure. This saves time on preparation and processing.

Since localization goes in parallel with the creation of the game, most often the material comes in parts, at best the publisher warns about the possibility of starting to voice certain characters. The primary task in this case is to identify the files of these characters among those sent, prepare them for recording, calculate the amount of time needed for recording, because this allows you to calculate the load of the studio or studios, if you write in several, prepare texts. If the publisher has not provided such information, it is better to clarify it.

Minimum information required before recording

Before you start recording, it is important to ask the publisher the right questions. They help to plan forces and resources correctly and set priorities.

  • What type of files are to be recorded?
  • Do I need lip sync?
  • Are there videos in the game and how many?
  • Technical requirements for audio files?
  • What is the deadline for submitting the material?

Types of audio files and calculation of the time required for recording:

  • lip sync (from “lip synchronization”), usually recorded under video (voice-over);
  • image sync is a close analogue of “lip sync”, the presence of video is mandatory (voice-over);
  • sound sync — synchronization under the wave (voice-over);
  • strict time constraint — it is necessary to strictly keep within the length of the audio file (“voice-over” and AI);
  • no time constraint — length limit +/-10% (AI).

There are standards of pronunciation in an hour, proven by our experience, which we adhere to when planning time for recording a certain part of the text.
Knowing the number of words of a certain type of record and the rate of recording per hour, you can calculate the amount of time needed to record a particular text.


As a conclusion, I would like to draw the following conclusion: in our opinion, the key to effective sound recording is the steady execution of the following points:

  • proper planning;
  • maximum automation of processes;
  • timely selection of actors for roles;
  • high-quality translation and styling of the translated text;
  • well-coordinated work of the whole team and professionalism of each individual.

Here, in general, is what kind of experience in sound localization I wanted to share with you. I hope you were interested.

Related news