The current signal-injection workflow stores the signal chunks entirely in memory before injecting them and writing out the final frame files. This approach becomes inefficient—and in some cases infeasible when generating many long-duration signals, since memory usage can grow extremely large.