MICROPHONE INPUT

(Real Time voice)

AUDIO LOADING

(Librosa)

RESAMPLING

(16 kHz)

NORMALIZATION &

MONO CONVERSION

WHISPER ASR MODEL

(Speech → English Text)

ASR TEXT CLEANING

(Lowercase, Normalize)

REFERENCE TEXT

(Ground Truth)

WER CALCULATION

(Compare ASR vs Reference)

(ASR Accuracy Metric)