Achieving English lip-sync for a large volume of cutscenes in the global release of Monster Hunter Wilds: improving efficiency through the use of the automated voice animation system “SGX”

Capcom's AAA title Monster Hunter Wilds, which has sold more than 11.8 million copies worldwide, adopted Speech Graphics' SGX automated voice animation system to achieve natural English lip-sync for its global release. The central challenge in developing the English version was converting the Japanese lip-sync adjustments into English in a short period with a small team, and SGX was selected to handle the vast number of cutscenes and lines.

SGX was used not only for human characters but also for beastfolk characters such as Palico and Moriber. With a common rig structure, the team generated and verified the English lip-sync, while the setup with Speech Graphics was based on the in-house rig, with the team specifying any behaviors that were missing. The initial setup took approximately 1.5 to 2 months.

The team automated the process of converting vocal information such as breathing and laughter into text and reflecting it in the lip-sync by developing SGX pipeline tools. Although verification initially involved extensive manual work, the tools reduced the time required to complete the analysis output for one cut to approximately 1 minute, allowing the work to proceed in parallel with other tasks.

Approximately 70% of the analysis results could be used as-is, and in some cases more than 80% could be used without modification, while other cases, such as meeting scenes, yielded only around 50% to 60% usable results. In one early sandship scene, SGX's results were applied almost entirely as generated. Without SGX, a substantial amount of work would have been required for English lip-sync alone, so English production began after development of the Japanese version had reached a stable phase. Wider adoption of multilingual support and improvements to features that infer emotion from vocal intonation are expected in the future.