This study mainly focuses on investigating the devoicing phenomenon happening in Indonesian language. There are nine words selected as the sample of the study: [akad], [jilid], [wujud], [kutub], [sebab], [tertib], [caleg], [dialog], and [warteg]. All these words involve voiced stop consonants including bilabial /b/, alveolar /d/, and velar /g/ at coda position (final word). Ten participants from different age and gender were selected as participants, they subsequently produced the words and the data of their vocalizations were transcribed and analyzed using Praat software. Praat was used to bring the data of pitch (frequency), duration, and intensity. The result of the data suggests that the participant experienced the phonological process called devoicing while pronouncing the words. The words which in the first place involve voiced stop consonants, turn into voiceless stop consonants /p/, /t/, /k/ respectively. To give phonological explanation of devoicing, generative phonology and optimality theory were used. Generative phonology was used to show the changing process from level of representation to the surface level of representation as well as to formulate the phonological rule for the devoicing process. Meanwhile, optimality theory was used to describe the input and output for each word. Keywords: devoicing; generative phonology; optimality theory; stop consonants