<p>This article reports on the validation of Fabla, a researcher-developed and university-hosted smartphone app that facilitates naturalistic and secure collection of participants’ spoken responses to researcher questions. Fabla was developed to meet the need for tools that (a) collect longitudinal qualitative data and (b) capture speech biomarkers from participants’ natural environments. This study put Fabla to its first empirical test using a repeated-measures experimental design in which participants (<i>n</i> = 87) completed a 1-week voice daily diary via the Fabla app, and an identical 1-week text-entry daily diary administered via Qualtrics, with diary method order counterbalanced and randomized. A preregistered analysis plan investigated (1) adherence, usability, and acceptability of Fabla, (2) concurrent validity of voice diaries (vs. text-entry diaries) by comparing linguistic features obtained via each diary method, and (3) differences in the strength of the association between linguistic features and their known psychological correlates when assessed by voice versus text-entry diary. Voice diaries yielded more than double the mean daily language volume (word count) compared to text-entry diaries and received high usability and acceptability ratings. Linguistic markers consistently associated with depression in prior research were significantly associated with depression symptoms when assessed via voice but not text-entry diaries, and the difference in correlation magnitude was significant. Word-count-adjusted linguistic patterns were highly correlated between diary methods, with statistically significant mean differences observed for some linguistic dimensions in the presence of these associations. Fabla is a promising tool for collecting high-quality speech data from participants’ naturalistic environments, overcoming multiple limitations of text-entry responding.</p>