Termin – Detailansicht
Bachelor-Vortrag: Real-Time Own Voice Estimation for Hearables Utilizing Multiple Microphone Signals
Donnerstag, 17. Juni 2021
Hearable devices employ several microphones to be used in telecommunication applications as well as for active noice cancellation. It is proposed that the signals generated by these microphones can be combined into an accurate and low-interference estimate of the user's own voice in real time by using digital beamforming algorithms, namely the so-called delay-and-sum and minimum variance distortionless response beamformers, especially in conjunction with additional filtering in the shape of a bandpass or a Wiener filter.
To test this, several input samples are recorded on a hearable device and then processed with said algorithms. It is found that it is indeed possible to generate such a signal using the proposed algorithms, improving the signal to noise ratio by several decibels while maintaining a low distortion of the speech signal.