Back to LAB main
Better Loc Next Time, Part 1: Intro to Audio Localization for Games
Olivier Deslandes, Senior Vice President of Audio & Speech Technology at SIDE (PTW’s premium audio supplier), has worked in voice production for most of his career. His hard-earned wisdom about all things video game audio cannot be understated.
Recently Olivier spoke about Audio Localization at GameSoundCon in LA, and we’re excited to share some of the key takeaways from his talk with you here.
Read on to learn where to start when considering localized audio for your game.
It isn’t a matter of if one should pursue localization into other languages, but rather which languages to target, and at what level: from in-game text all the way up to a fully voiced game.
The reasons to pursue localization are well documented. For one, players appreciate a game more if it is culturally relevant to their own experience. They’ll find that a game is more engaging if they can play it in their own language, as opposed to puzzling through a language different from their own. More languages also mean a wider audience, so the potential player base grows larger, which leads to more revenue.
Depending on the game’s genre and scope, it might make sense to simply stick to localized text. Going ahead with audio localization, on the other hand, is a more complex process—but one that pays off greatly for the added immersion it provides players.
Long before audio recording takes place, a comprehensive localization strategy should be defined. Determining which languages and regions to target will narrow down your game’s budget and timeline.
Once the game script is completed, it’s best practice to create a Loc Kit, containing not only the script but also information on tone, themes, setting, plot, characters—all the necessary data to be used throughout the localization process.
Then comes the crucial moment of choosing a Loc partner. There are single-language vendors (SLVs) and multi-language vendors (MLVs), each with their own risks and benefits. An SLV is a specialist in a given language, meaning their results should be more focused, and commonly less expensive. However, if you’re targeting multiple languages, it can be cumbersome to manage several discrete SLVs to keep the timeline and the quality consistent.
With an MLV, you’re essentially getting a managed collection of SLVs. The larger outsourcers in the industry can scale up quickly across regions and offer enhanced security, more tech-oriented solutions and processes, and peace of mind. The downside is that they will likely be more expensive than SLVs.
Regardless of which model you choose, there are four main stages of the audio loc process to be aware of: script localization, voice-over (VO) recordings in the original and target language(s), and finally, localization QA.
We come to arguably the most important part of the process: the translation, or more accurately localization, of the script itself. This stage is crucial because poor translations will introduce errors further down the pipeline and make the process avoidably harder.
Localization is much more than just translating dialogue: every bit of text the player sees must be processed, including names, dates, units of measurement, and abbreviations, appearing anywhere in the game’s HUD, maps, menus, and so on. What’s more, text must be properly culturalized to the target region or location to avoid offending the sensibilities of that culture, which can include religious matters, alcohol and drug use, the depiction of blood, etc. And finally, adaptation of text for length can be a major hurdle, along with the need to properly sync dialogue.
Simply put, the better the translation is prepared ahead of actual voice recording, the smoother the process becomes.
With voice production, the goal is always a well-translated script, executed by a well-directed cast. But the path to get there is complex.
Once the game design is locked, the character briefs are ready, and there is a final recording script for the original or source version of the game, the voice director is hired and briefed on the game mechanics and “feel” of the narrative.
Casting then takes place, and recording begins. Voice strings are edited and named as the source dialogue audio base is built, which will constitute the reference for localized recording—including how characterization unfolds throughout the narrative. In an ideal world, the source language should be finalized before starting audio localization so that there is a complete and locked set of assets. The original recordings can also provide important additional artistic direction for the localization studio to identify what works and what doesn’t. However, with large titles, this is not always possible, so the work must be done in parallel on a rolling basis.
Having information on characterization as early as possible helps the Loc team make these adjustments at casting time, because they will not have the luxury to play around with characterization in-studio. Localization recording always goes quicker than the original recording, not only because Loc doesn’t have to “create” the character tone or intensity, but also because it is rarely afforded the same time or budget as the original.
Thus, localized recording takes place, including any post-production (or editorial), F/X, and mastering as requested. The localized VO is delivered to be checked before integration. What’s returned is mirrored localized audio databases that are consistent with the source: the same folder structure, matching file naming, and matching file count, which is paramount for integration and LQA. If timelines are compressed, one can integrate “dry” files and proceed with linguistic testing to save some time. The dry files will eventually be swapped with the final ones (voices with effects, final videos, etc.).
Once the files are ready and the audio is integrated in-game, it’s up to LQA, or localization quality assurance, to do its job. LQA is the process of evaluating the translation and the voiceover in the game itself against that of the target language. Unlike the mechanical aspects of functionality, compatibility, and compliance, which only need to be checked once per platform, LQA must occur for each language.
Proper LQA involves native speakers of the target language who have access to debug kits and copies of the build. Ideally, the same company that handled the translation and VO would perform LQA, assuming they are equipped for this service. Planning is key, once again, as the LQA provider will need to prepare resources ahead of time. Ultimately, it comes down to triage: knowing which bugs must be addressed immediately and which can be deferred.
Understand up front that there is no such thing as a completely bug-free build. Typical problems include subtitles not displaying or displaying out of sync; translations not making sense in-game due to context; audio files not playing; audio playing in the wrong language; incorrect audio playing; and volume variance, among others.
Working with the same company that handled your translation and audio is always a good option, as they’ll be able to scale up with bigger resources, provide tech support, have plenty of kits ready, and deal more easily with emergency demands. Getting the results for all languages at the same time can also be very helpful.