How is Audio Localized for Games?

Localization

Sound Localization

Game voice-over localization is a complex, technical process. Success hinges on understanding all of its parts and the ability to anticipate potential challenges.

Projects requiring sound localization are generally received in advance.

Project timelines can range from 2 weeks to 6+ months, which gives us sufficient time to strategize, assess available resources, construct realistic timelines, and submit material on time.

It is essential to review all process components, including translator and editor availability, studio availability based on existing/additional projects, and available voice actors who are up to the challenge.

If the project scale is large and the deadline is on a not-so-distant horizon, recording at multiple studios simultaneously is an option to consider.

Workflow Organization and Team

Each of the specialists on our team fills a unique, integral role.

The project manager, often one of the editors, organizes workflow and identifies functional group roles, issues text to translators, receives and reviews translated text from other editors, informs the sound group of materials to be recorded, participates in casting voice talent, and oversees the effective execution of project deadlines.

The sound director oversees the in-studio recording process, and works directly with the voice actors and sound technician in the studio.
The audio engineer is responsible for the quantity and quality of incoming (English) audio files and the quality of the resulting localized audio files.
The sound technician works in-studio and assists the sound director throughout the voice-over recording process.
Our engineering team develops a toolkit to facilitate preparation of audio for recording and tracking the recorded material.
The testing team evaluates the quality of the recorded voice-over.

Preparing for Localization

Sound localization projects begin well before the voice-over process. Effective planning and strategy are essential to a seamless localization process.

The process commences when the publisher's project manager sends the source text for voice-over localization, which includes scripts (text for primary and secondary character speech).

Translators are the first to work on the source text, followed by editors, who sync the text with the original audio files.

Well-synchronized text within the audio file removes the distraction of having to edit and layer synchronized text during the recording process, thereby allotting the maximum amount of time for recording. A well-selected cast of actors will be able to quickly and efficiently navigate the casting and recording process.

The audio engineer analyzes the source files, quantifies them relative to the script, and prepares for sound recording (for example, by assigning temporary attributes of sequence number, date, and record type, which enables the program to automatically group files by character, actor, or recording type.

Diligent preparation allows for a maximally efficient sound recording process and minimizes the number of duplicate and defective files.

How is audio localized for games?

Localization Hurdles

Addressing Grammatical gender

– F*** you!

– Buzz off!

This translation example highlights an issue requiring developer input. This is due to differences in grammatical structure between source (English) and target languages, namely the presence of gendered nouns and lack thereof in English. This occurs most frequently in localizations of secondary (non-player) character text.

Say, for example, a character in the original (English) game yells the line "Screw you!" Knowing the gender of the addressee (the "you") affects how the phrase is translated in Italian, and may lead to an instinctual bias towards the masculine declension. However, an elegant solution may be a translation that employs neutral language to bypass grammatical gender. A more complex scenario arises when a hero's gender is subject to change. In such cases, each scenario must be treated individually to effectively evaluate whether grammatically neutral alternatives can be substituted or, if not, sent to developers to be split into 2 lines.

Voice Casting

As translators work on translating the text and resolving issues like those mentioned above, the studio receives source (English) audio files.

Work with sound begins. Localization of a major game intended for simultaneous multilingual release often runs parallel to its development.

This introduces a number of nuances to the sound localization process. For example, developers may introduce major changes to parts of the translated and scanned text and send an updated script requesting that we "figure out" what's new and in need of revision.

Recording inevitably begins with casting. Once the publisher completes voicing for the game's main characters, samples of the source (English) text are provided to guide the casting process.

Casting decisions are made one of two ways: live or virtually.

For live casting, actors are invited to our in-house studio to perform and record selected passages of in-game text. For reference, either specific samples, video, or images of mo-cap sessions are provided. Clients have the option to attend live casting and often participate virtually through Skype.

For virtual casting, the project manager and sound team put together a selection of sound samples from the voice talent database, featuring actors they see as the best fit for a given project. The database consists of sound samples from voice actors who have prior experience working with our games department. As a rule, 3+ actors are selected as candidates for the main character, but this is subject to change based on client demands, such as, for example, a virtual versus live casting process for leading roles within the game.

An important component of the casting process is compiling the right selection of samples to present the publisher to inform casting decisions. The key is providing samples that effectively capture tone and correspond with the original genre. This means that if a warrior in the original (English) lets out a battle cry, there must also be a battle cry in the target sample, and if an elder speaks in the original, he must be represented in the localized sample.

This process revealed the lack of suitable sound samples as a recurring challenge. We decided to address the issue by developing what we refer to as the “ideal” casting process. We selected content to effectively demonstrate the actor’s full range of game voicing capabilities in the span of a 30-minute recording session. This includes the most common character types across age groups, and all incoming voice actors are required to provide these samples in order to be quickly and effectively matched with an appropriate project. During this process, we also evaluate the actor’s flexibility in playing different characters and adapting his/her voice to different age ranges.

Working with Voice Actors

We are constantly in search of feedback on our work and always on the lookout for general reviews of projects involving voice acting both in and outside of the game industry. We often encounter complaints from audiences and players of hearing "the same voices everywhere."

We believe the lack of professional voiceover actors to be a general problem that affects the media industry. In addition to talent and "know how" these actors must also possess the ability to work quickly and well.

As I mentioned earlier, sound localization usually takes place in a short time and runs parallel to the creation of the game itself. The sound director therefore invites only those who can be trusted to contribute to a project’s success and have a track record of reliability.

We are constantly looking for new voices, and each new project attracts up to 15% new actors—that is, those whom we have yet to work with. In other words, every 6th actor in the new project is new to us. Why so few? Each new actor presents a certain risk and this ratio allows us to manage the risk and rerecord "unsuccessful roles".

Usually, we begin working with new actors by giving them episodic roles, or so-called "grenades" which can be "screamed" without enthusiasm and serve as a dead giveaway. If we see that a new actor works hard and has potential to develop, we make a note of it and invite the actor back to the studio for additional opportunities to develop and showcase their expertise.

Actors must work in the industry, be local to Moscow and capable of commuting to our studio for recording, which excludes the risk of inconsistent recording quality. If an actor/actress combines voice acting with shooting on video, there is always a possibility that he/she might leave for another city and that the character voice-over will need to be redone. we will have to rewrite the role, etc.

Not all talented actors will be suited to voice games, or for voice acting, period. Even within the pool of experienced voice talent, not all candidates have what it takes to excel in game arena. Voicing games requires actors to assume character and effectively channel a role in a short time, often relying exclusively on the source (English) audio in the absence of relevant visual content. Character portraits are typically only available for the primary characters.

Voice-over Process in Localization

Defining Efficient Voice-over Recording

To facilitate the voice-over recording process, the prepared text should be available on hand.

When I say "prepared", I mean that this text is not only properly translated into the target language, but is synchronized with the game audio. The editor/sync specialist is the last team member to handle the localized content, and is responsible for matching the length of the target text to that of the original. This is done in Excel, allowing control over line length in both languages.

The Voice-over Process

After the casting process is complete, actors are contacted to confirm availability and review detailed scheduling for those cast to voice main characters.

Recording is carried out in a program that is convenient for the client. For example, if submitted material needs to be exclusively in the ProTools program, adjustments must be made.

Our goal is to return the recorded material ready to be integrated into the game and in adherence to technical requirements and project deadlines.

Translation is carried out in the same program as that of the provided source material.

There are various translation programs that facilitate and expedite the translation process, such as MemoQ, Trados and others. Despite the existence of advanced tools, text is frequently sent in Excel. This program also effectively accommodates records management and preparation of files for recording. It provides a broad range of possibilities for the efficient organization of materials, tracking received and recorded files, and managing intermediate deadlines, and built-in Visual Basic tools to customize utilities to your precise needs.

Most importantly, the content you'll be working with needs to have "identifying marks" which are elements that can be used to systematically organize the material. For example, by creating a uniform file name structure. This strategy allows more time to be devoted to preparation and processing.

Because the localization process runs parallel to game development, material frequently arrives piecemeal, with the best-case scenario involving some direction from the publisher regarding which characters we might be able to start voicing or otherwise. The goal here is to locate these files within the source content, prepare them for recording, and calculate the time needed to record, which is used to determine the amount of studio time and/or space needed (if using more than one studio), and to prepare the necessary texts. If the publisher hasn't provided this information, be sure to pose clarifying questions.

Information Required Prior to Recording

Prior to sound recording it is essential to ask the publisher the right questions. These will inform an efficient planning strategy, optimize the allocation of resources, and formulate clear priorities.

What type of file needs to be recorded?
Is lip syncing required?
Does the game contain video clips? If yes, how many?
What are the technical requirements for the audio files?
What is the final deadline for transfer of materials?

Types of audio files and submission timelines:

Lip sync — typically recorded for video (voice-over)
Image sync — similar to lip sync, requires video content
Sound sync — voice-over
Strict time constraint — requires strict adherence to audio file length (voice-over and AI)
No time constraint — ±10% of source audio file length (AI)

Our experience has surfaced standards for the volume of spoken language per hour and we adhere to these norms when planning the time needed for a given text selection.

Knowing the wordcount for a particular recording and the rate of recording per hour allows us to calculate the time needed to record it.