First page Back Continue Last page Overview Graphics
Representing Speech Data
Common transform is the mel-scale
The mel-scale is a log scale
Models human perception
Divides the signal into frequency bands
Returns the log-energy for each frequency band