In this paper, we present Pyrlato, an innovative tool developed in Python for collecting acoustic data from YouTube. The development of this tool was motivated by the need to conveniently collect real-world spoken data. By executing this Python code, researchers can obtain a spoken corpus of specific words, syllables, constituents, and more. We illustrate the main steps of the execution to demonstrate how it works and how to use it. Additionally, we provide a complete example for reference, demonstrating how to customize Pyrlato according to specific requirements. Finally, we discuss the future developments we intend to cover for Pyrlato.