Voice & speech processing

We specialize in end-to-end voice & speech technology—from clinical signal analysis to production-grade mobile apps—centered on impaired, alaryngeal, and clinical voice. Our work blends modern signal processing, multi-objective (Pareto) optimization, and deep generative models to deliver reliable assessment, rehabilitation, and real-world monitoring solutions.

• Clinical analytics & screening: Robust extraction of objective voice markers (e.g., AVQI-style composites), pathology screening and progression tracking, PD screening from voice, and device-agnostic scoring with calibration across microphones and environments.

• Enhancement & reconstruction: Noise-robust enhancement for alaryngeal speech; flow-based/generative synthesis to restore intelligibility and speaker likeness; artifact suppression and quality optimization using multi-criteria loss design.

• Mobile wellness & telemedicine: Cross-platform apps for point-of-care voice checks, triage, and longitudinal self-monitoring; secure cloud/edge pipelines for real-time inference and clinician dashboards.

• Rehabilitation & prosthetic voice: Algorithms and interfaces supporting tracheoesophageal users, with feedback-driven training and personalization.