A body-worn, real-time speech and voice biofeedback device is described. Data from an acoustic microphone and piezoelectric sensor worn comfortably in a neckband are streamed to a digital signal processor and a small, mobile computer, altogether able to fit into a pocket for extended use. User laryngeal and spectral characteristics are determined from the combination of sensor inputs. Selected vocal characteristics (e.g., vocal intensity, shimmer, jitter, spectral output, and fundamental frequency) are analyzed in real-time to provide immediate user feedback via tactile or visual response to indicate speech production pathologies including reduced loudness, pitch instability, or other features. With minimal training, this feedback can be immediately acted upon by the wearer to adjust speech and voice production characteristics accordingly. In addition, all data from input sensors are collected and stored in the computer's memory for offline analyses of speech and voice production characteristics. Extended-use, large-sample data collection addresses issues in the extant literature including ecological validity, reactivity (e.g., the Hawthorne effect), small sample sizes, and unaccounted for individual differences. This work offers a realistic description of voice use and assesses a wide range of functional and organic clinical conditions that are known to affect speech production (e.g., Parkinson's disease).