Accent recognition in foreign language speech using Speech Accent Archive Dataset. A 2-layer CNN is used as a model.
A variety of audiosignal features is used.
An opportunity to remove silence fragments from inspected audio samples.
Flexible parameters adjustments.