Massachusetts Institute of Technology

Sensimetrics Corporation

Joseph G Desloge

Patrick M Zurek

Louis D Braida

Charlotte M Reed

An approach to audio processing aims to improve intelligibility by amplifying time segments of an input signal when the level of the signal falls below a long-term average level of the input signal, for instance, introducing a time-varying gain such that the signal level of the amplified segment matches the long-term average level.

Normalizing signal energy for speech in fluctuating noise

An immersive hearing loss and auditory prostheses simulator allows a person who listens through the simulation system to experience an actual shift in his or her thresholds for detecting ambient sounds, in a way that is similar to the shift in thresholds experienced by a hearing-impaired person. The simulator shifts the listener's thresholds while also processing the input signals for suprathreshold stimulation. With a controlled degree of auditory threshold shift with loudness recruitment, a hearing loss simulator is made valid and flexible.

System and method for immersive simulation of hearing loss and auditory prostheses

An adaptive multiple-tap frequency domain digital filter processes an input signal vector X from an plurality of spatially separated transducers that detect energy from a plurality of sources including a target energy source and at least one non-target energy source. The filter receives and processes the input signal vector X to attenuate noise from non-target sources and provides an output signal vector Y. Tap weights Wfor the filter are selected by first parameterizing each of the tap weights W, such that each of the tap weights Wis characterized by a vector of parameters, and the solving for each parameter of the vectorby minimizing the expected power of the array output signal Y. A robustness-control transformation is then applied to the vector p to provide a robust vectorwherein the robustness-control transformation identifies and reduces target canceling components of the vectorwhile preserving non-target canceling components. Finally, the weight vector indicative of the filter tap weights is formed as a function of the vector. Notably, the present invention separates the robustness constraining process from the beamforming power minimization, in contrast to prior art techniques which combine the robustness constraint into the beamforming power minimization. The present invention uses a direct and flexible robustness control mechanism to yield a beamformer that provides good performance and is robust to a wide variety of adverse conditions.

Growing community of inventors

Brookline, MA, United States of America

Joseph G Desloge