This work is an attempt to create a spatial work using the sound effect generative AI that learned the sound data of the 36 spatial audio works evala has released in the past. It generates “evala-like sound” in real-time and on a multichannel basis, using both sound (the whistle sound evala uses at the beginning of his work as a signature) and text (8 titles from the works used for learning are selected for each channel) as reference (prompt). This project is an experiment exploring a new form of archive that can permanently inherit and create spatial audio works even if the artist is absent. And this work is its first trial model.