Provided is an in-ear device and associated computational support system that leverages machine learning to interpret sensor data descriptive of one or more in-ear phenomena during subvocalization by the user. An electronic device can receive sensor data generated by at least one sensor at least partially positioned within an ear of a user, wherein the sensor data was generated by the at least one sensor concurrently with the user subvocalizing a subvocalized utterance. The electronic device can then process the sensor data with a machine-learned subvocalization interpretation model to generate an interpretation of the subvocalized utterance as an output of the machine-learned subvocalization interpretation model.

Disclosed herein, among other things, are systems and methods for person-to-person voice communication between ear-wearable devices. A method includes receiving, using a microphone of a first hearing device configured to be worn on or in an ear of the first user, a first acoustic own voice signal from the first user. The method further includes transmitting, from the first hearing device via a wireless connection to a second hearing device configured to be worn on or in an ear of a second user, a first audio packet based on the received first acoustic own voice signal. The method also includes receiving, at the second hearing device via the wireless connection, the first audio packet from the first hearing device, and playing, at the second hearing device using a second receiver, an output signal for the second user based on the first audio packet.

A hearing device includes a processor and a wireless transceiver operatively connected with the processor and configured to connect via a wireless link with a remote device and to exchange bi-directional audio packets with the remote device. The processor is configured to monitor an audio link quality, and adapt a transmit power of the hearing device as a function of the audio link quality.

A hearing device comprises A) a forward path, comprising a1) an input unit for providing a time-domain electric input signal as digital samples, a2) an analysis filter bank configured to provide a time-frequency representation of said electric input signal, a3) a signal processing unit for processing a signal of the forward path and providing a number of processed channel-signals, B) an onset detector configured to receive said time-domain electric input signal before entering said analysis filter bank, and to provide an onset control signal dependent on a current first order derivative of an envelope thereof, C) a level estimation unit for estimating a current level of said frequency sub-band signals, and comprising c1) a level adjustment unit configured to adjust the current levels of said frequency sub-band signals, and to control said level adjustment in dependence of said onset control signal. The invention may be used in audio devices, e.g. hearing aids.

A hearing device includes: an input module for provision of a first input signal; a processor configured to provide an electrical output signal based on the first input signal; a receiver configured to provide an audio output signal; and a controller comprising a speech intelligibility estimator configured to determine a speech intelligibility indicator indicative of speech intelligibility based on the first input signal, wherein the controller is configured to control the processor based on the speech intelligibility indicator; wherein the speech intelligibility estimator comprises a decomposition module configured to decompose the first input signal into a first representation of the first input signal in a frequency domain, wherein the first representation comprises one or more elements representative of the first input signal; and wherein the decomposition module comprises one or more characterization blocks for characterizing the one or more elements of the first representation in the frequency domain.

Apparatuses, systems, and techniques are presented to reduce noise in audio. In at least one embodiment, one or more neural networks are used to determine a noise signal in one or more speech signals.

A method operates a hearing system having a hearing device and modifies an input signal for the purpose of sound output to a user and, applies multiple algorithms with a respective potency, as a result of which a respective algorithm is applied with a present potency in a present situation. The hearing system recurrently receives a report from the user indicating that the user is dissatisfied with the sound output in the present situation. The hearing system has a database, containing multiple weights for each algorithm, to rate a change of the potency. If a report is received, each of the algorithms is rated using the weights to ascertain an individual-case relevance for each of the algorithms, to assess the effect of a change of the potency. Multiple individual-case relevances are combined to form a relevance value for each algorithm, the relevance values are compared with one another.

A partner microphone unit comprising a) a multitude microphones for picking up a sound from the environment providing corresponding electric input signals, each comprising a target signal component and a noise signal component; b) a multi-input unit noise reduction system for providing an estimate Ŝ of the target sound s comprising the person's voice and comprising a multi-input beamformer filtering unit coupled to said input units and configured to determine filter weights for providing a beamformed signal, wherein signal components from other directions than a direction of the target signal source are attenuated, whereas signal components from the direction of the target signal source are left un-attenuated; c) antenna and transceiver circuitry for establishing an audio link to another device; and wherein the multi-input beamformer filtering unit comprises an adaptive beamformer is provided.

A system, including a signal input, a processor a signal output, wherein the processor is configured to generate an instruction related to data related to a recipient of a sensory prosthesis based on input into the signal input, and the signal output is configured to output data indicative of the instruction.

A hearing aid comprises a sensor configured for detecting a focus of an end user on a real sound source, a microphone assembly configured for converting sounds into electrical signals, a speaker configured for converting the electrical signals into sounds, and a control subsystem configured for modifying the direction and/or distance of a greatest sensitivity of the microphone assembly based on detected focus. A virtual image generation system comprises memory storing a three-dimensional scene, a sensor configured for detecting a focus of the end user on a sound source, a speaker configured for conveying sounds to the end user, and a control subsystem configured for causing the speaker to preferentially convey a sound originating from the sound source in response to detection of the focus, and for rendering image frames of the scene, and a display subsystem configured for sequentially displaying the image frames to the end user.