import math
from . import base
[docs]class MeanVarianceNorm(base.OfflineComputation):
"""
Pre-processing step to normalize mean and variance.
frame = (frame - mean) / sqrt(variance)
Args:
mean (float): The mean to use for normalization.
variance (float): The variance to use for normalization.s
"""
def __init__(self, mean, variance, parent=None, name=None):
super(MeanVarianceNorm, self).__init__(parent=parent, name=name)
self.mean = mean
self.variance = variance
self.std = math.sqrt(variance)
def compute(self, frames, sampling_rate, corpus=None, utterance=None):
return (frames - self.mean) / self.std