Development of low-power, low memory, hybrid algorithms (digital signal processing + machine learning) for various speech processing tasks.
- The ideal candidates are on their last year of their graduate program (PhD or MS).
- Desired skills:
Experience with digital signal processing (DSP) as well as deep learning
Experience with MATLAB and Python.
Knowledge of one of the common machine learning training frameworks: PyTorch,
TensorFlow, Caffe
Familiar with common neural network architectures, namely
LSTMs, CNNs, RNNs, Generative networks, Sequence-to-sequence networks, etc.
Position 2
Vector quantized representation for speech generative networks and its application for voice user interface.
- Desired skills:
Fundamental deep learning theory, practice and good experience of using deep learning framework such as PyTorch in GPU powered Linux environment