Masterarbeit - Details

Enhanced Own Voice Perception for Hearables Using a Selective Feed-Forward Filter

Betreuer: Till Hardenbicker

Themengebiet: Active Noise Cancelling, Sensor Fusion, Bandbreitenerweiterung, Maschinelles Lernen

Werkzeuge: Matlab, Python


In recent years, wireless bluetooth earbuds finally arrived at the market for consumer electronics. These so called hearables have the potential to improve the hearing and listening experience of the wearer. The possibilities include selective listening, augmented reality, active noise cancellation and many more. However, these features require a multitude of different algorithms to be employed in real time in a computation and power constrained environment.

In our research, we focus on two yet opposing operation modes: While the so called Active Noise Cancellation (ANC) tries to cancel all environmental sounds by means of an additive signal with a complementary phase, the Occlusion Effect Cancellation (OEC) makes the device acoustically transparent, such that the own voice is perceived more naturally.

The task of this thesis is to develop a selective ANC mode, where environmental sounds are still blocked, except for the air conducted parts of the speakers own voice. In order to separate the own voice from other sound sources in the signal, the hearable’s inner (error) microphone signal might be of help, as it contains a low pass filtered version of the own voice with attenuated environmental sounds.

As part of this thesis, different approaches shall be researched and analyzed regarding their suitability to solve the adaptive filter problem. This can include machine learning approaches as well as efficient algorithms based on common signal processing. One or multiple promising approaches shall be implemented. Thereafter, their performance shall be evaluated systematically based on real world example signals. The analysis should reflect the importance of computational constraints of the hearable platform.