Abstract: Understanding and modeling emotions from speech is a fundamental challenge in speech processing and a key enabler of emotionally intelligent human-computer interaction. However, defining and ...
Abstract: This letter presents a new target speech recognition problem, where the target speech is defined by a keyword. For instance, when a person speaks “Hey Google” or “Help Me”, we hope the model ...
MMdnn is a set of tools to help users inter-operate among different deep learning frameworks. E.g. model conversion and visualization. Convert models between Caffe, Keras, MXNet, Tensorflow, CNTK, … ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results