voice-clinic-desktop workflow

voice-clinic-desktop workflow A workflow diagram generated by Archify. 01 / Preprocessing (process_recording) 02 / Decision (models loaded by sex) 03 / Output 04 / Training (prepare / evaluate / train) Sustained vowel · WAV · any sample rate · Preprocessing (process_recording) Sustained vowel WAV · any sample rate Load and denoise · resample · log-MMSE or 3 others · Preprocessing (process_recording) Load and denoise resample · log-MMSE or 3 others Voice activity detection · volume + ZCR · multi-segment · Preprocessing (process_recording) Voice activity detection volume + ZCR · multi-segment Cepstral features · 13 MFCC + Δ · voiced frames only · Preprocessing (process_recording) Cepstral features 13 MFCC + Δ · voiced frames only Model files · 2 sexes × 2 tasks · Decision (models loaded by sex) Model files 2 sexes × 2 tasks Screening model · normal vs pathological · Decision (models loaded by sex) · pooled frame scores + threshold Screening model normal vs pathological pooled frame scores + threshold Subtype model · scored only if pathological · Decision (models loaded by sex) Subtype model scored only if pathological Interface and report · waveform · spectrogram · scores · Output Interface and report waveform · spectrogram · scores dataset.csv · path · label · group · sex · Training (prepare / evaluate / train) dataset.csv path · label · group · sex prepare / train · same preprocessing · group k-fold · Training (prepare / evaluate / train) prepare / train same preprocessing · group k-fold voiced frames loaded by sex if normal if pathological 4 model files segments Legend Interface and report Preprocessing Models Model files Input data