Multi-Microphone Signal Processing and Machine Learning

点击翻译按钮获取中文摘要

In this project, we will address the problem of improving quality of the audio signals in environments withmultiple microphones and multiple speakers. In future smart houses, it is envisioned that there would be avariety of sensors including many inexpensive microphones connected wirelessly to a central station. Toimprove the quality of audio signals, data from multiple microphones will need to be combined using machinelearning and signal processing algorithms.One of the challenges that we will address in this project is to synchronize audio signals from each microphonewith low latency so that it appears to the user as if there is no delay in processing. Another challenge is to usemultiple microphones to supress noise and amplify the audio signals of interest.This project is beneficial to Canada because it has potential to improve future smart homes. In addition, Canadais investing significantly in machine learning and artificial intelligence, and these methods will be used in thisproject. The supporting company will benefit from the results of this project because the methods developed inthis research could be directly applied to their current products.