Research in cognitive neuroscience has traditionally relied on simple, highly controlled stimuli due to the difficulty in developing standardized, ecologically valid stimulus sets. However, recently, there has been a consensus that using ecologically valid stimuli is imperative to generalize results beyond controlled laboratory settings. The current study presents a naturalistic audio stimulus database that meets the specific requirements of cognitive neuroscience for short, recognizable, and emotionally rated stimuli. To create such a database, the current study collected 291 audio files from a wide range of sources. 361 participants rated the audio clips on emotionality, arousal, and recognizability, and subsequently freely described the audios by typing what they believed the sound to be. The text responses of the participants were embedded and clustered using an unsupervised machine-learning algorithm to derive a participant-grounded organization of auditory object categories. The results indicate that the participants were confident of their ability to recognize the audios, while emotionality and arousal ratings showed broad variability, enabling future researchers to select stimuli that best suit their experimental needs. Furthermore, the final database comprises ten distinct semantic categories, providing a diverse set of auditory stimuli. Going beyond its experimental utility, the current study offers a methodological framework that can be replicated to develop other types of stimulus sets.