Multiple-Instance, Cascaded Classification for Keyword Spotting in Narrow-Band Audio

2017/11/21 by AbdulKader, Ahmad, Nassar, Kareem, El-Geish, Mohamed +2
#Audio and Speech Processing (eess.AS) #Computation and Language (cs.CL) #FOS: Computer and information sciences #FOS: Electrical engineering #Machine Learning (cs.LG) #Sound (cs.SD) #electronic engineering #information engineering

paper · doi:10.48550/arxiv.1711.08058

Abstract

We propose using cascaded classifiers for a keyword spotting (KWS) task on narrow-band (NB), 8kHz audio acquired in non-IID environments -- a more challenging task than most state-of-the-art KWS systems face. We present a model that incorporates Deep Neural Networks (DNNs), cascading, multiple-feature representations, and multiple-instance learning. The cascaded classifiers handle the task's class imbalance and reduce power consumption on computationally-constrained devices via early termination. The KWS system achieves a false negative rate of 6% at an hourly false positive rate of 0.75

Related

• Multichannel Keyword Spotting for Noisy Conditions
• Adaptive Noise Resilient Keyword Spotting Using One-Shot Learning
• Small-Footprint Keyword Spotting on Raw Audio Data with\n Sinc-Convolutions
• DCCRN-KWS: an audio bias based model for noise robust small-footprint keyword spotting
• End-to-End Streaming Keyword Spotting