Livoa LogoLivoa
Audio Standardization


Sampling rate normalization (8–16 kHz)

Convert to Mono
Amplitude Normalization


Reduce variability between sessions

Light Pre-emphasis


Stabilize spectral tilt

Band-pass Filtering


100–1000 Hz or 50–2000 Hz

Windowing


Hann or Hamming window, overlapping frames

Short-Time Fourier Transform (STFT)
Mel Filterbank Projection


40–128 bands, Slaney-style normalization

Logarithmic Compression (log-Mel)


Stabilize dynamic range, remove outliers

Per-band Standardization


Z-score or cepstral mean/variance normalization

Temporal Deltas & Delta-Deltas (Optional)


Capture short-range dynamics

dd

by kkkkk

0
0 uses