ABSTRACT
Stuttering is a speech disorder that disrupts fluency and rhythm, requiring precise assessment and therapy. This research presents an automated system for stuttering assessment and therapy, utilizing advanced speech signal analysis and real-time visual feedback. Speech characteristics of two stutterers used for case study, aged 22 and 16 years, were analyzed using techniques such as time-domain waveforms, spectrograms, pitch contours, energy and zero-crossing rate (ZCR) analyses, formant analysis, and real-time energy feedback. The 22-year-old's signal spanned 70 seconds with amplitudes between -0.4 and 0.4, while the 16-year-old's signal lasted 140 seconds with amplitudes from -1 to 1. Spectrograms showed dominant frequencies below 5 kHz, with more continuous spectral energy for the younger stutterer. Pitch ranged between 50 Hz and 400 Hz for both, with smoother transitions in the 16-year-old. Energy analysis revealed maximum energy levels of 0.04 and 0.3 for the older patient, while both patients showed ZCR peaks of 0.5. Formant analysis indicated consistent frequency ranges for F1 (500–1000 Hz), F2 (1500–3000 Hz), and F3 (3500–4500 Hz). Real-time feedback revealed energy peaks of 7×104 units at 17 seconds for the older stutterer and 9×10³ units at 8 seconds for the younger stutterer. This system combines quantitative insights with personalized, real-time feedback, enhancing stuttering assessment accuracy and enabling adaptive therapy, offering a solution for stuttered speech management.
