Game Localization: Voice Recording vs. Glossary Translation — What Should Publishers Prioritize? 5 Things Teams Miss 3 Months Before Global Launch

For a game publisher PM three months out from global launch, the first question is usually "When do we start voice recording?" But teams that jump into recording without a finalized glossary often end up re-recording entire sessions over a single character name change, or face community confusion when item names vary across languages. In game localization, voice is the most visible element—but if that voice isn't built on consistent terminology, the entire localization can collapse.

Key Takeaways

Why Focusing Only on 'Voice' Leads to Localization Failure

Teams new to game localization typically identify "voice recording" as the biggest challenge. In-game voiceover is indeed expensive and time-consuming—from casting to recording to editing. But focusing solely on voice means overlooking the real problem: the text those voices carry can end up inconsistent across languages.

Consider a mobile RPG publisher that, two months before global launch, translated Korean source dialogue into English, Japanese, and Chinese, cast voice actors for each language, and completed recording. Then the marketing team requested a change to a character's English name spelling. The recorded pronunciation no longer matched the new spelling, forcing a full re-record of every line featuring that character. The result: budget overruns of 30%+ and a two-week launch delay.

This pattern repeats for one clear reason: voice production started before the glossary was finalized. In game localization, a glossary is a document that standardizes character names, items, skills, locations, and core world-building terms across languages. Without it, each translator and recording session will render the same term differently, and unifying them later means tearing everything apart and starting over.

Industry-standard guides warn against this repeatedly. For example, [IGDA's Best Practices for Game Localization](https://igda-website.s3.us-east-2.amazonaws.com/wp-content/uploads/2021/04/09142137/Best-Practices-for-Game-Localization-v22.pdf) emphasizes that for games with audio, recording must begin only after script translation and glossary approval, and that a translation lead should be present during recording sessions.

Why Glossary Translation Comes First: Games Live on 'World Consistency'

Game localization differs from film or TV dubbing. A movie ends in two hours, but a game repeats the same terms hundreds of times over dozens of hours of play. Players search for item names, debate skill combos in communities, and watch strategy videos while internalizing the game's world. If the same item is called "Sword of Fire" in the UI, "Flame Sword" in quest text, and "Fire Sword" in voice lines, players are confused and immersion breaks.

In practice, game translation specialists treat "terminology consistency" as the first quality metric. One global game localization studio reported that "with a glossary, translating 10,000 lines is 40% faster and revision requests drop by half." A glossary eliminates the need for translators to deliberate over every word—they simply apply the approved term.

This workflow is also emphasized in [Microsoft's official game localization guide](https://learn.microsoft.com/es-es/globalization/localization/localize-games). According to that guide, three months before global launch, teams must complete source dialogue finalization, glossary and style guide approval, and change-freeze criteria before scheduling voice recording.

Glossaries fall into three main categories:

For RPG or strategy games with heavy proper-noun usage, starting localization without a glossary is inherently risky. One indie studio preparing a 30-language launch finalized glossaries first, enabling each language's translators and voice teams to work from the same baseline—resulting in almost zero terminology bugs during QA.

5 Must-Check Items Before Voice Recording

Before scheduling voice recording for global launch, check these five items. Missing even one will lead to rework costs and schedule delays.

1. Is the Glossary Finalized for All Languages?

First, confirm that glossaries exist for every target language. If only the Korean source glossary is ready and English/Japanese versions are missing, each language's translators will define terms independently. A glossary isn't just a word-mapping table—it should include pronunciation guides (e.g., stress placement in character names), contextual usage examples, and forbidden expressions. Dubbing services like Polita automatically prioritize glossary terms during AI dubbing generation when you upload the glossary, and expert reviewers refine pronunciation and intonation to glossary standards during the polishing stage.

2. Are Voice Scripts Synchronized with In-Game Text?

Voice scripts are usually managed separately, but if they don't match actual in-game UI and quest text, players will notice "the screen says A but the voice says B." Especially when patches or updates modify text, if voice isn't updated in parallel, version mismatches accumulate. Before voice production, lock the final build text and establish a workflow to immediately share any subsequent changes with the voice team.

3. Have You Reviewed Cultural Taboos and Local Regulations for Each Language?

Game localization is cultural adaptation, not just translation. For example, China is sensitive to certain historical and political terms, the Middle East requires avoiding religious expressions, and Germany legally prohibits certain symbols. If these issues surface after voice recording, you'll need to rewrite entire lines—or, in the worst case, abandon that market. Native reviewers for each language should flag cultural and regulatory risks starting at the glossary stage.

4. Are Voice Actor Character Briefs and Tone Guides Ready?

Game voice isn't just reading lines—it's expressing character personality, emotion, and context. But overseas voice actors often haven't played the original game, and recording from script alone can result in a tone that diverges from the source. Especially when recording multiple languages simultaneously, if each language's actor interprets the same character differently, global players will complain that "this character's personality changes by language." Before recording, prepare per-character tone guides (e.g., "bright and upbeat," "cynical but warm underneath") and reference audio, then review initial recording samples to align direction.

5. Are Revision/Re-Recording Processes and Budget Secured?

No matter how well you prepare, integrating voice into the actual game and playtesting will reveal areas needing revision. Dialogue timing may not match cutscene length, certain word pronunciations may sound awkward, or bugs may emerge during QA. You won't need to re-record everything, but you must have a process for selectively revising specific lines. Polita offers up to two free revisions on standard plans, and because work is glossary-based, when only certain terms change, only those lines can be quickly regenerated. Without this flexibility, even small fixes take weeks and incur extra costs.

In the AI Dubbing Era, Glossaries Have Become Even More Critical

The game industry is increasingly turning to AI dubbing. Traditional recording costs hundreds of thousands of won per language and takes weeks, but AI dubbing can cut costs by more than half and reduce turnaround to days. In fact, global game localization firms like Altagram are actively adopting AI technology, emphasizing 30-language audio services.

But AI dubbing doesn't eliminate the need for glossaries. In fact, in the AI era, glossaries are more important than ever. AI generates voice based on training data and input scripts—if the same term is used inconsistently across contexts, AI will produce inconsistent pronunciation and intonation. For example, if the character name "Aria" is romanized as "Aria" in some lines but "Eria" in others, AI will treat them as different words and pronounce them differently.

The real-world quality and limitations of AI dubbing are also addressed in [University of Malta's research on AI dubbing quality](https://www.um.edu.mt/library/oar/bitstream/123456789/107156/1/Quality_in_interlingual_AI_dubbing.pdf). According to that study, AI-based script adaptation accuracy is around 70%, with remaining errors primarily stemming from translation and synchronization issues. This proves that even with AI dubbing, human translator and language director review, plus glossary-based polishing, remain essential.

Polita solves this problem with a glossary-based workflow. When clients upload a glossary, the AI dubbing generation stage prioritizes those terms and produces consistent pronunciation. Then, during the expert review stage, human reviewers finalize term pronunciation and intonation, with selective edits as needed. One mobile game publisher dubbed over 10,000 in-game lines in five languages through Polita, and thanks to the glossary, character names and skill names were pronounced consistently across all lines.

Additionally, AI dubbing enables fast revisions. Traditional recording requires re-booking voice actors and studio time, but AI dubbing can regenerate specific lines within hours after script edits. This is especially advantageous for live-service games with frequent updates. Every time a new character or event is added, voice can be updated quickly, and adding new terms to the glossary allows expansion while maintaining tone consistency with existing voice.

Practical Checklist: Localization Priorities 3 Months Before Global Launch

For teams preparing game localization for the first time, here's a priority-ordered timeline starting three months before global launch.

3 Months Before Launch (D-90)

2 Months Before Launch (D-60)

1 Month Before Launch (D-30)

2 Weeks Before Launch (D-14)

The key in this timeline is glossary first, voice later. Without a finalized glossary, every subsequent stage becomes unstable, and revision costs snowball.

This phased prioritization and risk management is also emphasized in [Keywords Studios' game localization guide](https://www.keywordsstudios.com/en/about-us/news-events/news/a-step-by-step-guide-to-game-localization/). The broader the scope of full dubbing work—glossary, style guide, script adaptation, casting, audio engineering, lip-sync review—the more critical it is to lock terms and standards early for overall schedule and quality.

Why Polita Is Well-Suited for Game Localization

When game publishers choose a dubbing partner, the three most important criteria are: ability to maintain terminology consistency, fast revision turnaround, and reasonable cost. Polita meets all three as a B2B dubbing service.

First, Polita provides glossary-based workflows as standard. When clients upload a glossary, the AI dubbing generation stage prioritizes those terms, and the expert review stage has human reviewers finalize pronunciation and intonation. One mobile game dubbed over 10,000 lines through Polita, and character names and skill names were pronounced consistently across all lines.

Second, Polita enables fast revisions. Standard plans include up to two free revisions, and when terms change, only the affected lines can be selectively regenerated. When game updates or patches add dialogue, new lines can be added quickly while maintaining existing voice tone. Standard turnaround is approximately 5 business days for a 10-minute video, with AI dubbing generation completed in 1–2 days.

Third, pricing is reasonable. Polita's Standard plan is $12–15 per minute per language—less than half the cost of traditional recording. The Pro plan is $20–25 per minute per language and offers voice conversion technology that replicates the original actor's timing, breathing, and emotion. There's no minimum order, so even a single trailer or small update can be commissioned without burden.

Finally, Polita supports 70+ languages and has processed over 45,000 videos annually. With experience delivering large-scale multilingual projects across industries—games, animation, OTT, education—Polita is well-versed in complex terminology management and simultaneous multi-language execution. NDA-based security workflows are also in place, enabling safe handling of sensitive content like unreleased games.

---

Voice is undeniably important in game localization. But if that voice isn't built on correct terminology, even the best actors' recordings will confuse players and break immersion. Right now, as you prepare for global launch, the first priority isn't a flashy trailer voiceover—it's the quiet glossary that supports every language. A solid glossary makes voice production, updates, and community communication flow naturally. Conversely, starting with voice before the glossary guarantees a nightmare of tearing everything apart right before launch.

Frequently Asked Questions

Q. Who should create the glossary, and when?

The glossary should be drafted in Korean by the game planning team or localization PM, then translated and reviewed by native translators for each target language. It must be finalized at least 3 months before global launch to avoid delays in subsequent text translation and voice production. The glossary should include not only character, item, and skill names, but also pronunciation guides and contextual usage examples.

Q. Does AI dubbing really need a glossary?

Yes. Without a glossary, AI dubbing can pronounce the same term differently across contexts. Services like Polita prioritize glossary terms during AI generation when you upload one, and expert reviewers finalize pronunciation and intonation during the polishing stage to ensure consistency. A glossary lets you enjoy AI dubbing's speed and cost advantages while maintaining quality.

Q. What happens if terms change after voice recording?

Traditional recording requires re-booking voice actors for re-recording, but AI dubbing can regenerate specific lines within hours after script edits. Polita offers up to two free revisions on standard plans, and because work is glossary-based, when only certain terms change, only those lines can be selectively revised.

Q. How much does game localization cost?

Cost depends on the number of target languages, volume of dialogue, and voice quality tier. For Polita, the Standard plan is $12–15 per minute per language, and the Pro plan is $20–25 per minute per language. Dubbing a 10-minute game trailer in 5 languages costs approximately $600–750 on Standard—less than half the cost of traditional recording, with no minimum order, so even small projects can start without burden.

Q. What should I watch out for when working on multiple languages simultaneously?

Finalize glossaries for each language first, and ensure all language translators and voice production teams share the same glossaries. Especially for character names and proper nouns, transliteration methods vary by language, so pronunciation guides must be clear. Polita supports 70+ languages and provides workflows that maintain terminology consistency even during simultaneous multi-language execution.

References

- IGDA's Best Practices for Game Localization — Emphasizes that for games with audio, recording must begin only after script translation and glossary approval, and that a translation lead should be present during recording sessions - Microsoft's official game localization guide — States that three months before global launch, teams must complete source dialogue finalization, glossary and style guide approval, and change-freeze criteria before scheduling voice recording - Keywords Studios' game localization guide — Emphasizes that full dubbing includes glossary, style guide, script adaptation, casting, audio engineering, lip-sync review, and that locking terms and standards early is critical for overall schedule and quality - University of Malta's research on AI dubbing quality — Proves that AI-based script adaptation accuracy is around 70%, with remaining errors primarily stemming from translation and synchronization issues, and that human translator and language director review, plus glossary-based polishing, remain essential