The Arizona Child Acoustic Database consists of longitudinal audio recordings from a group of children over a critical period of growth and development (ages 2-7 years). The goal of this database is to 1) document acoustic changes in speech production that may be related to physical growth 2) inform development of a model of speech production for child talkers. This work was funded by NSF BSC-1145011 awarded to Kate Bunton, Ph.D. and Brad Story, Ph.D, Principal Investigators. This database contains longitudinal audio recordings of 55 American English speaking children between the ages of 2-7 at 3-month intervals. Since children began the study at different ages, some children have fewer recording sessions than others. The database can also be used to provide cross-sectional data for children of a specific age. Please refer to the subject data table for information on specific sessions available here http://arizona.openrepository.com/arizona/handle/10150/316065. All children were recorded using the same protocol; therefore, task numbers are consistent across children and sessions. A calibration tone is included as Record 1 for all sessions. The speech protocol focused on production of English monopthong and diphthong vowels in isolation, sVd, hVd, and monosyllabic real words. In addition, the protocol includes several nonsense vowel-to-vowel transitions. Speakers were prompted either verbally by investigators or by graphical prompts. Details of the protocol with reference to task numbers can be found in the protocol spreadsheet available here http://arizona.openrepository.com/arizona/handle/10150/316065. Details on data recording: All samples were recorded digitally using an AKG SE 300B microphone with a mouth to mic distance of approximately 10 inches. Signals were recorded digitally using a Marantz PMD671, 16 bit PCM (uncompressed) at 44.1KHz. Recordings are made available in .wav format. Individual zip files contain all recordings from a single session.