We design a voice-controlled video game to integrate Catalan into gaming using speech technologies developed under the Aina project. The game is designed to elicit natural speech commands from players. However, a significant challenge in this endeavor is the limited availability of Catalan-language Spoken Language Understanding (SLU) datasets, especially those covering specialized linguistic domains relevant to interactive gaming environments. To address this, we implement a cascading SLU system that combines automatic speech recognition (ASR) with roBERTa-based models previously trained in Catalan. The latter was finetuned as a multi-task classifier by generating synthetic transcriptions from a small set of human-written examples. With acceptable accuracy and time inference, our goal is to evaluate its performance in-game and gather feedback from users.