2.0.3

• Introduced a new model based on Deep Neural Networks (DNN) - Yamnet VAD, which provides audio classifier for 521 different type of sound;
• Split libraries to different modules;
• Fixed issue in WEBRTC VAD.